Published online before print November 28, 2013, doi: 10.3174/ajnr.A3809
AJNR 2013 34: E128-E129

L.C. Hygino da Cruz Jr
Clinicas CDPI and IRM
Federal University of Rio de Janeiro
Rio de Janeiro, Brazil

I read with great interest the review article by Hock et al1 and agree with the authors that ¹H-MR spectroscopy can be of great value in the assessment of a variety of spinal cord diseases. The difficulty in determining a specific diagnosis of these lesions is a challenge in their management. It is necessary to predict an appropriate treatment and avoid an unnecessary invasive intervention, such as biopsy or surgery. Thus, a reliable differential diagnosis can be obtained by using functional MR imaging techniques, which have been used for a long time to evaluate brain lesions. Technical limitations due to a low signal-to-noise ratio and the proximity to the bone and CSF interface have been described as important limitations to the use of functional MR images in the evaluation of spinal cord lesions.

We describe a case of a 79-year-old man with systemic non-Hodgkin lymphoma, with a solitary, solid, enhancing lesion in the cervical spine (Fig 1). Single-voxel ¹H-MR spectroscopy was performed in a 1.5T MR imaging clinical scanner (Fig 2). Saturation bands, to reduce CSF pulsation artifacts and to suppress signals from outside the voxel, were placed surrounding the voxel. We did not use electrocardiography or respiratory triggering or gating. The voxel size used was 10 × 10 × 20 mm3. TE was 30 ms and TR was 1500 ms; the number of averages was 160, resulting in a total scanning time of 4 minutes.

To our knowledge and on the basis of the extended review of the literature shown by Hock et al,1 no previous publication described the MR spectroscopy changes observed in a spinal cord lymphoma. Intramedullary spinal lymphoma is a very rare neoplasm that accounts for approximately 3.3% of all CNS lymphomas, which consist of only 1% of all lymphomas in the body.2 Most are solitary lesions located in the cervical spine.

Previous reports used ¹H-MR spectroscopy to evaluate brain lymphoma and basically found elevated lipid peaks, a high choline/creatine ratio, and a low N-acetylaspartate/creatine ratio.3 ¹H-MR spectroscopy findings of spinal lesions were similar to those described in the brain. In this particular case, we also performed diffusion-weighted MR imaging. The lesion demonstrated restricted diffusion, which has been described as related to the high cellularity and a high nuclear-cytoplasmic ratio, characteristic findings of lymphoma.3 On the basis of these findings and with the clinical history of the patient, we made a presumptive diagnosis of spinal cord lymphoma, avoiding an invasive diagnostic intervention.

We demonstrated the feasibility and the potential of applying MR spectroscopy to assess spinal cord lesions in daily clinical practice, even though some technical obstacles can be observed. We used a clinical 1.5T MR imaging scanner to obtain reliable MR spectra and in less than the average scanning time presented by Hock et al1; this scanning time, in our view, can be reliably used in clinical practice. Thus, one may conclude that it is not totally necessary to perform the examination in a 3T MR imaging scanner with a long scanning time.

In conclusion, we propose that by exploiting the higher SNR available with new 1.5T scanners with multichannel coils, MR spectra can be obtained in a 1.5T scanner with a scanning time as short as 4 minutes, without compromising quality. The clinical accuracy of this technique needs to be investigated in a broader clinical setting in patients with spinal cord lesions.

References

1. Hock A, Henning A, Boesiger P, et al. ¹H-MR spectroscopy in the human spinal cord. AJNR Am J Neuroradiol 2013;34:1682–89 » Abstract/FREE Full Text
2. Koeller KK, Rosenblum RS, Morrison AL. Neoplasms of the spinal cord and filum terminale: radiologic-pathologic correlation. Radiographics 2000;20:1721–49 » Abstract/FREE Full Text
3. Haldorsen IS, Espeland A, Larsson EM. Central nervous system lymphoma: characteristic findings on traditional and advanced imaging. AJNR Am J Neuroradiol 2011;32:984–92 » Abstract/FREE Full Text

Reply

Published online before print November 28, 2013, doi: 10.3174/ajnr.A3810
AJNR 2013 34: E130

A. Hock^a,b
1Institute for Biomedical Engineering
University and ETH Zurich
Zurich, Switzerland
^bDepartment of Psychiatry, Psychotherapy, and Psychosomatics
Zurich University Hospital for Psychiatry
Zurich, Switzerland

A. Henning^c,d
cInstitute for Biomedical Engineering
University and ETH Zurich
Zurich, Switzerland
^dMax Planck Institute for Biological Cybernetics
Tübingen, Germany

S.S. Kollias^e
eInstitute of Neuroradiology
University Hospital of Zurich
Zurich, Switzerland

We thank Dr Luiz Celso Hygino Cruz Jr for his interest in our review article1 on spinal cord MR spectroscopy, and we appreciate the presentation of his own spinal cord MR spectroscopy data acquired from a patient with a non-Hodgkin lymphoma. It is a good example of the feasibility and usefulness of MR spectroscopy in the spinal cord and demonstrates that additional information, which is complementary to other imaging methods, can be obtained. The presentation of these findings could help other clinical groups during the differentiation process in similar cases and may stimulate interest in using spinal cord MR spectroscopy in clinical routine. We also agree that the acquisition protocol can be simplified in comparison with our suggestions in the cited review article1 in this specific case.

However, the simplification of the acquisition protocol is not generally applicable but was enabled by the specific conditions in a case of a space-occupying lesion filling the spinal canal as reported by Dr Hygino da Cruz. The presented lesion is significantly larger compared with the normal diameter of a healthy or even atrophic spinal cord, thus a larger voxel size can be used. In Dr Cruz’s article, the voxel size was 2 mL compared with approximately 1.2 mL in healthy spinal cord,2 which implies an SNR increase by a factor of approximately 1.7. To achieve the same SNR in a smaller spinal cord MR spectroscopy voxel from a healthy or atrophic spinal cord, an increase of at least a factor of 2.9 in scanning time is necessary. In addition, the space-occupying lesion hinders the CSF flow and thus reduces the need for flow- and motion-correction methods. The application at 1.5T in conjunction with a larger lesion also reduces susceptibility changes and thus lowers the requirements for B0 shimming methodology. Compared with 3T, measurements at 1.5T theoretically have a lower SNR. In practice, the SNR difference between 1.5T and 3T is not substantial,3 due to a better B0 homogeneity at 1.5T leading to a smaller absolute line width along with a more accurate flip angle calibration in the presence of a more homogeneous transmit B1 field and negligible chemical shift displacement artifacts at 1.5T.

Furthermore, some lesions show a dramatic increase of specific metabolite concentrations (eg, the presented case seems to exhibit an elevated Cho peak), which can be easily observed with a reduced SNR (unfortunately, the SNR value of the measurement was not reported). The particular situation (a space-occupying lesion versus a normal-diameter spinal cord) minimizes the technical challenges of MR spectroscopy and makes the simplified spinal cord MR spectroscopy protocol used by Dr Cruz and coworkers applicable to the reported case.

In conclusion, in some cases, especially when the voxel size can be increased (eg, in space-occupying lesions), a reduction of the scanning time is possible. However, a simplified and shortened MR spectroscopy protocol applied to measurements in other spinal cord conditions might lead to bad spectral quality and potentially erroneous conclusions. Therefore, an MR spectroscopy scan protocol leading to high-quality MR spectroscopy data across different spinal cord diseases is indispensable.

Nevertheless, we congratulate Dr Cruz for his report and are looking forward to seeing more detailed reports of spinal cord MR spectroscopy in various conditions, including quality indicators like SNR and line width.

References

1. Hock A, Henning A, Boesiger P, et al. ¹H-MR spectroscopy in the human spinal cord. AJNR Am J Neuroradiol 2013;34:1682–89 » Abstract/FREE Full Text
2. Hock A, MacMillan EL, Fuchs A, et al. Non-water-suppressed proton MR spectroscopy improves spectral quality in the human spinal cord. Magn Reson Med 2013;69:1253–60 » CrossRef » Medline
3. Callot V, Le Fur Y, Ranjeva JP, et al. Spinal cord diffusion tensor imaging (DTI) and ¹H-MR spectroscopy (MRS) at 1.5 T and 3 T. In: Proceedings of the 18th Annual Meeting of the International Society for Magnetic Resonance in Medicine, Stockholm, Sweden. May 1–7, 2010:2453 » Search Google Scholar