Neuroimaging in Sensory Neuronopathy Raphael Fernandes Casseb, Alberto Rolim Muro Martinez, Jean Levi Ribeiro de Paiva, Marcondes Cavalcante Franc¸a Junior From the Department of Neurology and Neuroimaging Laboratory–School of Medicine, University of Campinas, Campinas, Brazil.
ABSTRACT Sensory neuronopathies (SN) are a group of disorders characterized by primary damage to the dorsal root ganglia neurons. Clinical features include multifocal areas of hypoaesthesia, pain, dysautonomia, and sensory ataxia, which is the major source of disability. Diagnosis relies upon clinical assessment and nerve conductions studies, but sometimes it is difficult to distinguish SN from similar conditions, such as axonal polyneuropathies and some myelopathies. In this scenario, underdiagnosis is certainly an important issue for SN patients and additional diagnostic tools are needed. MRI is able to evaluate the dorsal columns of the spinal cord and has proven useful in the workup of SN patients. Although T2 weighted hyperintensity restricted to the posterior fasciculi without contrast enhancement is the typical finding, additional abnormalities have been recently reported. The aim of this review is to gather available information on neuroimaging findings of SN, discuss their clinical correlates and the potential impact of novel MRI-based techniques. Keywords: Sensory neuronopathy, MRI, posterior cord, dorsal root ganglia, spinal cord. Acceptance: Received July 17, 2014, and in revised form October 18, 2014. Accepted for publication November 27, 2014. Correspondence: Address correspondence to Marcondes C. Franc¸a Jr, MD, PhD, Assistant professor, Department of Neurology, Universidade of Campinas–UNICAMP, Rua Tessalia ´ Vieira de Camargo, 126, Cidade Universitaria “Zeferino Vaz,” Campinas, SP, Brazil–13083-970. E-mail: [email protected] Financial Support: Fundac¸ao ˜ de Amparo a` Pesquisa do Estado de Sao ˜ Paulo (FAPESP). Grant: 2013/01766–7. J Neuroimaging 2015;25:704-709. DOI: 10.1111/jon.12210
Introduction Sensory neuronopathies (SN) are a rare, but distinct subgroup of peripheral neuropathies, characterized by primary damage to the cell body of the dorsal root ganglia (DRG) neurons, which leads to degeneration of its central and peripheral projections.1 SN usually present as a patchy, asymmetrical, non-length dependent sensory loss with preserved strength. Proprioceptive sensory loss and ataxia (indicating damage to large neurons) are the most common clinical features and the main cause of disability. Degeneration of small and medium neurons can also happen, leading to positive sensory symptoms, like pain and hyperaesthesia. Dysautonomia can also be found, particularly in immune-mediated and paraneoplastic SN. The etiology is diverse, such as paraneoplastic, immune-mediated, infective, iatrogenic, and vitamin related. Despite that, around 50% of the patients present an idiopathic SN.2 SN patients can present either subacute or chronic onset, depending on the etiology. In the former, the diagnosis is easier, due to its nonlength dependent sensory loss with preserved motor functions. On the other hand, slowly-evolving SN often resembles chronic axonal polyneuropathies, which makes the differential diagnosis challenging.3 In this scenario, underdiagnosis is an important problem concerning the SN, and novel diagnostic tools are certainly needed. Early recognition of SN is important in clinical practice because it narrows the list of possible etiologies in comparison to polyneuropathies. In addition, it enables early detection of potentially severe diseases which are known to be associated with SN, such as neoplasias. Furthermore, prompt diagnosis would allow treatment to be instituted sooner for those types
704
Copyright
of treatable SN. This is particularly important because damage in SN is located in the cell bodies of sensory neurons, and this results in smaller chance of recovering compared to the polyneuropathies, which affect distal axons. The gold standard diagnostic method for SN is histological analysis of the DRG. However it is an invasive and technically difficult procedure,4 which limits its use to a very few qualified institutions. In fact, the diagnosis still relies upon clinical suspicion and abnormalities in nerve conduction studies (NCS). NCS shows a global, rather than distal, decrease in sensory nerve action potential amplitudes, with preserved motor NCS and needle EMG.5,6 Methods that enable detection of damage or dysfunction to the central projections of the DRG neurons are also very informative. There are previous reports on the use of somatosensory evoked potentials in SN, but this technique lacks sensitivity.7 This is due to extensive and often severe compromise of peripheral sensory fibers in SN. Spinal cord MRI has proven useful in such difficult cases because it can assess central projections independently of peripheral impairment. Several studies have indeed shown that patients with SN present dorsal lesions7–16 and others also showed cord atrophy.14,15 In addition, these abnormalities might be useful as prognostic markers in chronic SN.12 Despite that, some important questions about the role of neuroimaging in the workup of SN remain unclear, such as the sensitivity of the method, and the time of dorsal hyperintensity signal onset. Therefore, we attempted to address these issues by reviewing available data on the use of MRI in SN. We also discussed the typical MRI findings, as well as the potential role of novel
◦ 2015 by the American Society of Neuroimaging C
Fig 1. Cervical spinal cord T2 weighted image. Sagittal and axial slices from two patients with hs-T2WI and a healthy control. The hyperintense signal is highlighted by the arrows in the sagittal images. Patient in the middle (B and E) presents more intense symptoms in his legs, and hs-T2WI is restricted to gracillis fasciculus. Patient on the left (A and D) presents widespread symptoms, and hs-T2WI is diffuse. Control subject on the right (C and F) shows no abnormal signal. imaging techniques to assist in the diagnosis, follow-up, and even therapy of these patients.
Table 1. Frequency of hs-T2WI in SN Patients in Different Studies Study
Context 8
Okumura et al were possibly the first group to approach SN with MRI. They performed necropsy of a patient with SN and removed the whole spinal cord for further MRI and histological analysis. They found hyperintense T2 signal in the posterior columns of the spinal cord, with severe histological degeneration in the same region. Many other studies highlighted this abnormal hyperintensity,7–16 specially seen on T210,11,13–15 or T2*9,12,16 weighted images, and, therefore, the next section is dedicated to it.
Hyperintense Signal in T2 or T2* Weighted Spinal Cord Image (hs-T2WI) Different groups investigated SN with MRI, and this is the most reported finding. T2 or T2* weighted images of the spinal cord show a hyperintense signal in the posterior columns (Fig 1), which is caused by DRG neuronal death and secondary degeneration of its central projections. Damaged projections are later substituted by gliotic tissue, which has higher water concentration than normal CNS and causes the spinal hs-T2WI. According to Franca et al14 it extends for at least five cervical vertebrae in most cases, and Sghirlanzoni et al3 stated that it is detected all over the spinal cord.
Signal Importance SN diagnosis is not always an easy task, because other diseases may also present ataxia as their main symptom. Differentiation from cerebellar ataxias is well achieved by history and examination. However, it is not so straightforward when it involves other demyelinating neuropathies.17 Camdessanche´ et al7 proposed a score form to ease diagnosis, allowing to categorize patients in three levels: possible, probable or definite SN. Definite diagnosis is seldom established, because it demands DRG biopsy. Therefore, most patients are classified
Lauria et al (2000)11 Mori et al (2001)12 Takahashi et al (2003)27 Mori et al (2005)16 Damasceno et al (2008)13 Camdessanche´ et al (2009)7 Bao et al (2013)15 Average
Patients with hs-T2WI*
Total Patients
Frequency (%)
22 12 3 9 14 1 5
29 14 4 12 18 22 9
75.9 85.7 75.0 75.0 77.8 4.5 55.6 75 ± 28
*T2 weighted hyperintense signal at the posterior columns.
as probable SN, and hyperintense T2 signal is one of the major criteria to establish this diagnostic category. However, even when definite diagnosis is not achieved, effective management of the disease is still attainable.
Signal Onset There is still no consensus about the moment of signal onset. Possibly, the most substantial information available is from Sghirlanzoni et al.3 They refer to an AIDS subacute SN patient from a previous study11 in which he/she was scanned twice: once at the beginning of symptoms with no hs-T2WI present, and 6 months later, when hs-T2WI was already present. However, factors such as the aetiology and the rate of disease progression may influence the time of onset, thus justifying the need for further investigations. A related issue, concerning time and signal spreading, was pointed out by Waragai et al,10 during the investigation of a chronic idiopathic SN case, in which was verified hs-T2WI presence exclusively at lumbar level at first MRI examination, but 2 years later it was also detected at cervical level.
Signal Presence/Absence Frequency of hs-T2WI signal is variable in available studies, ranging between 4.5% and 85.7% of all SN patients (Table 1).
Casseb et al: Neuroimaging in Sensory Neuronopathy
705
Overall, hs-T2WI appears to be a sensitive sign; such variation might be attributed to different inclusion criteria and disease duration of the patients enrolled in each of the studies. It is possible that different etiologies also influence the appearance of this MRI sign, but none of the studies specifically addressed this issue. In contrast to the usual large-fiber SN, patients with smallfiber SN infrequently present hs-T2WI as reported by Gorson et al.18
Specificity Hs-T2WI is not a specific SN hallmark, since other spinal diseases may show this same feature. Okumura et al8 mentioned several conditions causing spinal cord degeneration. However, only a smaller group has already been shown to produce the hs-T2WI: subacute combined spinal cord degeneration (SCSCD; deficiency of vitamin B12—Fig 2), deficiency of vitamin E11 and Copper,14 and HIV-related vacuolar myelopathy.19 In SCSCD, differentiation may be achieved by analysis of the hs-T2WI before and after vitamin therapy, because the hs-T2WI tends to disappear after treatment: Hemmer et al20 reported that 2 of their 4 patients with vitamin B12 deficiency presented hs-T2WI, which disappeared after treatment. This probably happens because myelin rather than axonal damage underlies SCSCD. Hs-T2WI in the lateral columns is another useful clue to differentiate SCSCD and SN, because it is often found in the former, but not in the latter. Occasionally, autoimmune diseases may generate hs-T2WI in the posterior cord as well, but classically there is gadolinium enhancement, and this is not a characteristic of the SN hs-T2WI.
Clinical Correlates of hs-T2WI Some studies attempted to investigate the correlation between clinical features and the distribution of hs-T2WI. Mori et al12 described MRI findings in a series of Sjogren’s syndrome pa¨ tients with SN. In those subjects with both cuneatus and gracilis fasciculi hs-T2WI, there were severe and diffuse sensory deficits (involving limbs and trunk) and also dysautonomia. In contrast, when this hyperintensity was restricted to the fasciculus gracilis, patients presented mild or moderate symptoms that were limited to the limbs. An additional insight from dorsal columns involvement in SN is an etiological one, seen in AIDSrelated SN when hs-T2WI is classically restricted to the gracillis tract.3 Such selective pattern however is not exclusive for AIDSrelated SN; we have recently seen a seronegative patient with SN that presented an identical finding (Fig 1E).
Image Quality Detection of hs-T2WI is largely dependent on the image quality. This is particularly important because image boundaries are often not clear cut in spine MRI, which makes delimitation of dorsal column and neighbor structures sometimes difficult. In our experience, this abnormal signal is more easily identified in axial, rather than sagittal images. Bao et al,15 for instance, compared T2* images with multipleecho image combination (MEDIC), a specific type of T2* weighted image, and found the latter technique more sensitive to detect abnormalities. Indeed, they found hs-T2WI in 5 out of 9 patients with traditional T2* weighted image, and in 8 out of the 9 same patients with MEDIC. 706
Fig 2. Cervical spinal cord T2 weighted image. Sagittal (top) and axial (bottom) slices of a SCSCD patient. The red arrow (top image) highlights the hyperintense signal in the spinal cord as found in the SN case. We also see a hyperintense signal in the axial image.
Reproducibility Most studies relied upon visual analysis to identify hs-T2WI, and very few assessed how reproducible it was between different evaluators. It is rather probable that such analyses could benefit from a standardized/automatic process, as it would remove personal bias. Mori et al12 were able to divide the signal into moderate and high intensity groups, suggesting that a computerized method could be useful for quantification. Bao et al15 did measure the signal intensity from a circular region in the dorsal column of axial T2* images, compared values between patients and controls and found significant differences. However, they did not mention the criterion to decide if the signal was hyperintense, nor if it was performed automatically.
Additional Image Findings Atrophy Spinal Cord
By analyzing axial spinal cord T1 images with an in-house software, Franca et al14 showed significant cervical spinal cord atrophy in patients with chronic SN (Fig 3). Interestingly, atrophy correlated with clinical disability and with the extent of neurophysiological abnormalities. Bao et al15 found similar results with MEDIC images. They compared SN patients with a disease control group (motor neuron disease and subacute combined degeneration) and with normal controls. Significant atrophy in the spinal cord and reduction in the normalized anteroposterior dimension of the spine were verified exclusively for SN patients at the level of C7. They also found that anteroposterior and laterolateral
Journal of Neuroimaging Vol 25 No 5 September/October 2015
Fig 3. Layout of the SpineSeg software. Axial T1 weighted image revealing cervical spinal cord atrophy at C3 level in a SN patient (A), compared to a healthy control (B). Blue line represents the ellipse fitted to the spinal cord (in yellow).
Fig 4. Midline sagittal T1 weighted images of the cerebellum showing two patients with upper vermian atrophy (A and B), and a healthy control (C). normalized dimensions of the thoracic spine were reduced in SN patients. These previous studies were performed in a cohort of SN patients with mixed etiologies. It is not yet clear whether there is a correlation between etiology and extent of cord atrophy. To address this issue, quantitative studies with a single-etiology group of patients would be important. Recently, Chevis et al21 performed such a study in a large cohort of Friedreich’s ataxia patients, which is the prototypical genetic SN. Authors demonstrated cord atrophy and flattening as well as a direct correlation between cord area reduction and clinical disability. Disease duration also correlated with cord atrophy in these patients. Cerebellum
Cerebellar atrophy has been considered a rare finding in SN patients, but it is possibly overlooked. Damasceno et al22 used a semiautomated method to measure volumes of the whole cerebellum (CEV) as well as the anterior lobe of the cerebellum (ALV). These values were then normalized to whole brain volume and compared between healthy controls and patients. Authors failed to identify group differences, but some patients had extremely reduced volumes (smaller than 2.5 SDs of the control group mean values). This suggests that cerebellar atrophy is not common, but may be found in occasional SN patients (Fig 4).
Dorsal Root Ganglion
DRG are very small structures, so that most MRI acquisition protocols fail to image them properly. Recently Bao et al15 were able to visualize DRG using MEDIC images. They then modeled DRG at C7 level as an ellipse and showed that its normalized axes (“length” and “width” of the DRG) were reduced in SN patients, when compared to healthy control and disease control groups. Although promising, these results need to be replicated before widespread use.
Other Findings Some MRI-based studies performed in the early stages of SN demonstrated alterations outside the dorsal columns. T2 weighted signal changes were observed in a patient with Sjogren’s syndrome on the medial lemniscus and also in the ¨ pons and cerebellar peduncles with a normal cervical image, reflecting proximal sensory pathways impairment and a more diffuse immune-mediated attack.23 Bao et al15 demonstrated high signal intensity in cervical nerve roots without changes in dorsal columns in a patient with recent onset of SN, and in a follow-up exam 7 months after SN onset, showed hs-T2WI in dorsal columns and decreased high signal intensity. Bao et al15 also attempted to quantify DRG signal intensity at C7 on MEDIC images. They found significantly higher intensity in SN patients compared to both healthy and disease controls. In contrast, when using TIRM images (turbo inversion
Casseb et al: Neuroimaging in Sensory Neuronopathy
707
recovery magnitude sequence), no difference was found in signal intensity from C7 nerve roots, but the diameter was smaller in SN patient group.
Final Remarks Spinal cord MRI should not be considered in isolation to assist in the diagnosis of SN. One must take into account clinical and neurophysiological data. However, there are challenging patients who may benefit from detailed MRI analysis, such as those presenting purely sensory abnormalities but with a worsening gradient toward the distal legs. MRI is widely available and noninvasive procedure in contrast to the techniques that rely on biopsy analysis. We thus believe that MRI should be considered as an additional tool in the work-up of SN patients.
Future Perspectives We somehow understand basic hs-T2WI characteristics, but some of them are still unclear. Future work should address specific questions, such as the time of hs-T2WI onset and whether there is any etiology-specific pattern of abnormalities. To accomplish that, longitudinal investigations would be necessary, and it would require assessment of patients in the early stages of the disease (when there is no hs-T2WI) until the hs-T2WI is detected. The major issue is to find patients at early stages, since they are normally brought into medical attention in advanced stage, making it difficult to perform such studies. Neuroimaging progress is uncovering new fields to be explored. MR diffusion tensor imaging (DTI), for instance, is being employed in the investigation of other spinal diseases. Some interesting results were reported for amyotrophic lateral sclerosis showing reduction of fractional anisotropy and increased ADC values in the lateral columns of the cervical spinal cord.24 Since SN also involves degeneration of the long spinal tracts, we hypothesize that cord DTI might also prove useful to study SN patients. Moreover, higher field MR equipment may also favor investigation, since image quality (spatial resolution and signalto-noise ratio) is deeply impacted. In line with that, a recent study with an ALS patient, using a 7T MR system, showed impressive images of corticospinal tract damage at cervical spinal cord.25 Most of the aforementioned papers did not focus on brain MRI abnormalities. Nonetheless, cerebellar damage has been already reported in a few patients. We believe that encephalic involvement in SN has been overlooked and therefore, this is certainly an area to be further explored. Functional magnetic resonance imaging (fMRI) is another promising technique to investigate neuronal behavior both during mental rest and task conditions. Weeks et al26 investigated 3 SN patients with absent proprioception using fMRI and PET scans, and verified that during finger movements, patients did show cerebellar activation both in fMRI and PET examinations, even with absent proprioception sense. This favored the hypothesis that cerebellum coordinates voluntary movement through the integration of multiple inputs and not only the sensory ones. Overall, this study emphasizes that SN is a useful human model to study sensory-motor integration through MRI-based techniques. In the long term, such studies might be able to identify relevant targets in the brain amenable
708
for therapeutic interventions (eg, neuromodulation) in SN patients. The authors thank CAPES and FAPESP for funding this project, and Abril Jimenez for the language revision.
References 1. Denny-Brown D. Primary sensory neuropathy with muscular changes associated with carcinoma. J Neurol Neurosurg Psychiatry 1948;11:73-87. 2. Martinez AR, Nunes MB, Nucci A, et al. Sensory neuronopathy and autoimmune diseases. Autoimmune Dis 2012;2012:873587. 3. Sghirlanzoni A, Pareyson D, Lauria G. Sensory neuron diseases. Lancet Neurol 2005;4:349-61. 4. Colli BO, Carlotti CG Jr, Assirati JA Jr, et al. Dorsal root ganglionectomy for the diagnosis of sensory neuropathies. Surgical technique and results. Surg Neurol 2008;69:266-73. 5. Garcia RU, Ricardo JA, Horta CA, et al. Ulnar sensory-motor amplitude ratio: a new tool to differentiate ganglionopathy from polyneuropathy. Arq Neuropsiquiatr 2013;71:465-9. 6. Lauria G, Pareyson D, Sghirlanzoni A. Neurophysiological diagnosis of acquired sensory ganglionopathies. Eur Neurol 2003;50:14652. 7. Camdessanche JP, Jousserand G, Ferraud K, et al. The pattern and diagnostic criteria of sensory neuronopathy: a case-control study. Brain 2009;132:1723-33. 8. Okumura R, Asato R, Shimada T, et al. Degeneration of the posterior columns of the spinal cord: postmortem MRI and histopathology. J Comput Assist Tomogr 1992;16:865-7. 9. Sobue G, Yasuda T, Kumazawa K, et al. MRI demonstrates dorsal column involvement of the spinal cord in Sjogren’s syndromeassociated neuropathy. Neurology 1995;45:592-3. 10. Waragai M, Takaya Y, Hayashi M. Serial MR findings of chronic idiopathic ataxic neuropathy. J Neurol Sci 1997;151:93-5. 11. Lauria G, Pareyson D, Grisoli M, et al. Clinical and magnetic resonance imaging findings in chronic sensory ganglionopathies. Ann Neurol 2000;47:104-9. 12. Mori K, Koike H, Misu K, et al. Spinal cord magnetic resonance imaging demonstrates sensory neuronal involvement and clinical severity in neuronopathy associated with Sjogren’s syndrome. J Neurol Neurosurg Psychiatry 2001;71:488-92. 13. Damasceno A, Franca MC Jr, Nucci A. Chronic acquired sensory neuron diseases. Eur J Neurol 2008;15:1400-5. 14. Franca MC Jr, D’Abreu A, Zanardi VA, et al. MRI shows dorsal lesions and spinal cord atrophy in chronic sensory neuronopathies. J Neuroimaging 2008;18:168-72. 15. Bao YF, Tang WJ, Zhu DQ, et al. Sensory neuronopathy involves the spinal cord and brachial plexus: a quantitative study employing multiple-echo data image combination (MEDIC) and turbo inversion recovery magnitude (TIRM). Neuroradiology 2013;55:41-8. 16. Mori K, Iijima M, Koike H, et al. The wide spectrum of clinical manifestations in Sjogren’s syndrome-associated neuropathy. Brain 2005;128:2518-34. 17. Sheikh SI, Amato AA. The dorsal root ganglion under attack: the acquired sensory ganglionopathies. Pract Neurol 2010;10:326-34. 18. Gorson KC, Herrmann DN, Thiagarajan R, et al. Non-length dependent small fibre neuropathy/ganglionopathy. J Neurol Neurosurg Psychiatry 2008;79:163-9. 19. Chong J, Di Rocco A, Tagliati M, et al. MR findings in AIDSassociated myelopathy. AJNR Am J Neuroradiol 1999;20:1412-6. 20. Hemmer B, Glocker FX, Schumacher M, et al. Subacute combined degeneration: clinical, electrophysiological, and magnetic resonance imaging findings. J Neurol Neurosurg Psychiatry 1998;65:822-7. 21. Chevis CF, da Silva CB, D’Abreu A, et al. Spinal cord atrophy correlates with disability in Friedreich’s ataxia. Cerebellum 2013;12:43-7.
Journal of Neuroimaging Vol 25 No 5 September/October 2015
22. Damasceno A, Franca MC Jr, Cendes F, et al. Cerebellar atrophy is infrequently [corrected] associated with non-paraneoplastic sensory neuronopathy. Arq Neuropsiquiatr 2011;69:602-6. 23. Damasceno A, Franca MC Jr, Zanardi VA, et al. Brainstem involvement in Sjogren’s syndrome-related sensory neuronopathy. J Neuroimaging 2010;20:397-9. 24. Wang Y, Liu L, Ma L, et al. Preliminary study on cervical spinal cord in patients with amyotrophic lateral sclerosis using MR diffusion tensor imaging. Acad Radiol 2014;21:590-6.
25. Cohen-Adad J, Zhao W, Keil B, et al. 7-T MRI of the spinal cord can detect lateral corticospinal tract abnormality in amyotrophic lateral sclerosis. Muscle Nerve 2013;47:760-2. 26. Weeks RA, Gerloff C, Honda M, et al. Movement-related cerebellar activation in the absence of sensory input. J Neurophysiol 1999;82:484-8. 27. Takahashi Y, Takata T, Hoshino M, et al. Benefit of IVIG for longstanding ataxic sensory neuronopathy with Sjogren’s syndrome. IV immunoglobulin. Neurology 2003;60:503-5.
Casseb et al: Neuroimaging in Sensory Neuronopathy
709
ABSTRACT Sensory neuronopathies (SN) are a group of disorders characterized by primary damage to the dorsal root ganglia neurons. Clinical features include multifocal areas of hypoaesthesia, pain, dysautonomia, and sensory ataxia, which is the major source of disability. Diagnosis relies upon clinical assessment and nerve conductions studies, but sometimes it is difficult to distinguish SN from similar conditions, such as axonal polyneuropathies and some myelopathies. In this scenario, underdiagnosis is certainly an important issue for SN patients and additional diagnostic tools are needed. MRI is able to evaluate the dorsal columns of the spinal cord and has proven useful in the workup of SN patients. Although T2 weighted hyperintensity restricted to the posterior fasciculi without contrast enhancement is the typical finding, additional abnormalities have been recently reported. The aim of this review is to gather available information on neuroimaging findings of SN, discuss their clinical correlates and the potential impact of novel MRI-based techniques. Keywords: Sensory neuronopathy, MRI, posterior cord, dorsal root ganglia, spinal cord. Acceptance: Received July 17, 2014, and in revised form October 18, 2014. Accepted for publication November 27, 2014. Correspondence: Address correspondence to Marcondes C. Franc¸a Jr, MD, PhD, Assistant professor, Department of Neurology, Universidade of Campinas–UNICAMP, Rua Tessalia ´ Vieira de Camargo, 126, Cidade Universitaria “Zeferino Vaz,” Campinas, SP, Brazil–13083-970. E-mail: [email protected] Financial Support: Fundac¸ao ˜ de Amparo a` Pesquisa do Estado de Sao ˜ Paulo (FAPESP). Grant: 2013/01766–7. J Neuroimaging 2015;25:704-709. DOI: 10.1111/jon.12210
Introduction Sensory neuronopathies (SN) are a rare, but distinct subgroup of peripheral neuropathies, characterized by primary damage to the cell body of the dorsal root ganglia (DRG) neurons, which leads to degeneration of its central and peripheral projections.1 SN usually present as a patchy, asymmetrical, non-length dependent sensory loss with preserved strength. Proprioceptive sensory loss and ataxia (indicating damage to large neurons) are the most common clinical features and the main cause of disability. Degeneration of small and medium neurons can also happen, leading to positive sensory symptoms, like pain and hyperaesthesia. Dysautonomia can also be found, particularly in immune-mediated and paraneoplastic SN. The etiology is diverse, such as paraneoplastic, immune-mediated, infective, iatrogenic, and vitamin related. Despite that, around 50% of the patients present an idiopathic SN.2 SN patients can present either subacute or chronic onset, depending on the etiology. In the former, the diagnosis is easier, due to its nonlength dependent sensory loss with preserved motor functions. On the other hand, slowly-evolving SN often resembles chronic axonal polyneuropathies, which makes the differential diagnosis challenging.3 In this scenario, underdiagnosis is an important problem concerning the SN, and novel diagnostic tools are certainly needed. Early recognition of SN is important in clinical practice because it narrows the list of possible etiologies in comparison to polyneuropathies. In addition, it enables early detection of potentially severe diseases which are known to be associated with SN, such as neoplasias. Furthermore, prompt diagnosis would allow treatment to be instituted sooner for those types
704
Copyright
of treatable SN. This is particularly important because damage in SN is located in the cell bodies of sensory neurons, and this results in smaller chance of recovering compared to the polyneuropathies, which affect distal axons. The gold standard diagnostic method for SN is histological analysis of the DRG. However it is an invasive and technically difficult procedure,4 which limits its use to a very few qualified institutions. In fact, the diagnosis still relies upon clinical suspicion and abnormalities in nerve conduction studies (NCS). NCS shows a global, rather than distal, decrease in sensory nerve action potential amplitudes, with preserved motor NCS and needle EMG.5,6 Methods that enable detection of damage or dysfunction to the central projections of the DRG neurons are also very informative. There are previous reports on the use of somatosensory evoked potentials in SN, but this technique lacks sensitivity.7 This is due to extensive and often severe compromise of peripheral sensory fibers in SN. Spinal cord MRI has proven useful in such difficult cases because it can assess central projections independently of peripheral impairment. Several studies have indeed shown that patients with SN present dorsal lesions7–16 and others also showed cord atrophy.14,15 In addition, these abnormalities might be useful as prognostic markers in chronic SN.12 Despite that, some important questions about the role of neuroimaging in the workup of SN remain unclear, such as the sensitivity of the method, and the time of dorsal hyperintensity signal onset. Therefore, we attempted to address these issues by reviewing available data on the use of MRI in SN. We also discussed the typical MRI findings, as well as the potential role of novel
◦ 2015 by the American Society of Neuroimaging C
Fig 1. Cervical spinal cord T2 weighted image. Sagittal and axial slices from two patients with hs-T2WI and a healthy control. The hyperintense signal is highlighted by the arrows in the sagittal images. Patient in the middle (B and E) presents more intense symptoms in his legs, and hs-T2WI is restricted to gracillis fasciculus. Patient on the left (A and D) presents widespread symptoms, and hs-T2WI is diffuse. Control subject on the right (C and F) shows no abnormal signal. imaging techniques to assist in the diagnosis, follow-up, and even therapy of these patients.
Table 1. Frequency of hs-T2WI in SN Patients in Different Studies Study
Context 8
Okumura et al were possibly the first group to approach SN with MRI. They performed necropsy of a patient with SN and removed the whole spinal cord for further MRI and histological analysis. They found hyperintense T2 signal in the posterior columns of the spinal cord, with severe histological degeneration in the same region. Many other studies highlighted this abnormal hyperintensity,7–16 specially seen on T210,11,13–15 or T2*9,12,16 weighted images, and, therefore, the next section is dedicated to it.
Hyperintense Signal in T2 or T2* Weighted Spinal Cord Image (hs-T2WI) Different groups investigated SN with MRI, and this is the most reported finding. T2 or T2* weighted images of the spinal cord show a hyperintense signal in the posterior columns (Fig 1), which is caused by DRG neuronal death and secondary degeneration of its central projections. Damaged projections are later substituted by gliotic tissue, which has higher water concentration than normal CNS and causes the spinal hs-T2WI. According to Franca et al14 it extends for at least five cervical vertebrae in most cases, and Sghirlanzoni et al3 stated that it is detected all over the spinal cord.
Signal Importance SN diagnosis is not always an easy task, because other diseases may also present ataxia as their main symptom. Differentiation from cerebellar ataxias is well achieved by history and examination. However, it is not so straightforward when it involves other demyelinating neuropathies.17 Camdessanche´ et al7 proposed a score form to ease diagnosis, allowing to categorize patients in three levels: possible, probable or definite SN. Definite diagnosis is seldom established, because it demands DRG biopsy. Therefore, most patients are classified
Lauria et al (2000)11 Mori et al (2001)12 Takahashi et al (2003)27 Mori et al (2005)16 Damasceno et al (2008)13 Camdessanche´ et al (2009)7 Bao et al (2013)15 Average
Patients with hs-T2WI*
Total Patients
Frequency (%)
22 12 3 9 14 1 5
29 14 4 12 18 22 9
75.9 85.7 75.0 75.0 77.8 4.5 55.6 75 ± 28
*T2 weighted hyperintense signal at the posterior columns.
as probable SN, and hyperintense T2 signal is one of the major criteria to establish this diagnostic category. However, even when definite diagnosis is not achieved, effective management of the disease is still attainable.
Signal Onset There is still no consensus about the moment of signal onset. Possibly, the most substantial information available is from Sghirlanzoni et al.3 They refer to an AIDS subacute SN patient from a previous study11 in which he/she was scanned twice: once at the beginning of symptoms with no hs-T2WI present, and 6 months later, when hs-T2WI was already present. However, factors such as the aetiology and the rate of disease progression may influence the time of onset, thus justifying the need for further investigations. A related issue, concerning time and signal spreading, was pointed out by Waragai et al,10 during the investigation of a chronic idiopathic SN case, in which was verified hs-T2WI presence exclusively at lumbar level at first MRI examination, but 2 years later it was also detected at cervical level.
Signal Presence/Absence Frequency of hs-T2WI signal is variable in available studies, ranging between 4.5% and 85.7% of all SN patients (Table 1).
Casseb et al: Neuroimaging in Sensory Neuronopathy
705
Overall, hs-T2WI appears to be a sensitive sign; such variation might be attributed to different inclusion criteria and disease duration of the patients enrolled in each of the studies. It is possible that different etiologies also influence the appearance of this MRI sign, but none of the studies specifically addressed this issue. In contrast to the usual large-fiber SN, patients with smallfiber SN infrequently present hs-T2WI as reported by Gorson et al.18
Specificity Hs-T2WI is not a specific SN hallmark, since other spinal diseases may show this same feature. Okumura et al8 mentioned several conditions causing spinal cord degeneration. However, only a smaller group has already been shown to produce the hs-T2WI: subacute combined spinal cord degeneration (SCSCD; deficiency of vitamin B12—Fig 2), deficiency of vitamin E11 and Copper,14 and HIV-related vacuolar myelopathy.19 In SCSCD, differentiation may be achieved by analysis of the hs-T2WI before and after vitamin therapy, because the hs-T2WI tends to disappear after treatment: Hemmer et al20 reported that 2 of their 4 patients with vitamin B12 deficiency presented hs-T2WI, which disappeared after treatment. This probably happens because myelin rather than axonal damage underlies SCSCD. Hs-T2WI in the lateral columns is another useful clue to differentiate SCSCD and SN, because it is often found in the former, but not in the latter. Occasionally, autoimmune diseases may generate hs-T2WI in the posterior cord as well, but classically there is gadolinium enhancement, and this is not a characteristic of the SN hs-T2WI.
Clinical Correlates of hs-T2WI Some studies attempted to investigate the correlation between clinical features and the distribution of hs-T2WI. Mori et al12 described MRI findings in a series of Sjogren’s syndrome pa¨ tients with SN. In those subjects with both cuneatus and gracilis fasciculi hs-T2WI, there were severe and diffuse sensory deficits (involving limbs and trunk) and also dysautonomia. In contrast, when this hyperintensity was restricted to the fasciculus gracilis, patients presented mild or moderate symptoms that were limited to the limbs. An additional insight from dorsal columns involvement in SN is an etiological one, seen in AIDSrelated SN when hs-T2WI is classically restricted to the gracillis tract.3 Such selective pattern however is not exclusive for AIDSrelated SN; we have recently seen a seronegative patient with SN that presented an identical finding (Fig 1E).
Image Quality Detection of hs-T2WI is largely dependent on the image quality. This is particularly important because image boundaries are often not clear cut in spine MRI, which makes delimitation of dorsal column and neighbor structures sometimes difficult. In our experience, this abnormal signal is more easily identified in axial, rather than sagittal images. Bao et al,15 for instance, compared T2* images with multipleecho image combination (MEDIC), a specific type of T2* weighted image, and found the latter technique more sensitive to detect abnormalities. Indeed, they found hs-T2WI in 5 out of 9 patients with traditional T2* weighted image, and in 8 out of the 9 same patients with MEDIC. 706
Fig 2. Cervical spinal cord T2 weighted image. Sagittal (top) and axial (bottom) slices of a SCSCD patient. The red arrow (top image) highlights the hyperintense signal in the spinal cord as found in the SN case. We also see a hyperintense signal in the axial image.
Reproducibility Most studies relied upon visual analysis to identify hs-T2WI, and very few assessed how reproducible it was between different evaluators. It is rather probable that such analyses could benefit from a standardized/automatic process, as it would remove personal bias. Mori et al12 were able to divide the signal into moderate and high intensity groups, suggesting that a computerized method could be useful for quantification. Bao et al15 did measure the signal intensity from a circular region in the dorsal column of axial T2* images, compared values between patients and controls and found significant differences. However, they did not mention the criterion to decide if the signal was hyperintense, nor if it was performed automatically.
Additional Image Findings Atrophy Spinal Cord
By analyzing axial spinal cord T1 images with an in-house software, Franca et al14 showed significant cervical spinal cord atrophy in patients with chronic SN (Fig 3). Interestingly, atrophy correlated with clinical disability and with the extent of neurophysiological abnormalities. Bao et al15 found similar results with MEDIC images. They compared SN patients with a disease control group (motor neuron disease and subacute combined degeneration) and with normal controls. Significant atrophy in the spinal cord and reduction in the normalized anteroposterior dimension of the spine were verified exclusively for SN patients at the level of C7. They also found that anteroposterior and laterolateral
Journal of Neuroimaging Vol 25 No 5 September/October 2015
Fig 3. Layout of the SpineSeg software. Axial T1 weighted image revealing cervical spinal cord atrophy at C3 level in a SN patient (A), compared to a healthy control (B). Blue line represents the ellipse fitted to the spinal cord (in yellow).
Fig 4. Midline sagittal T1 weighted images of the cerebellum showing two patients with upper vermian atrophy (A and B), and a healthy control (C). normalized dimensions of the thoracic spine were reduced in SN patients. These previous studies were performed in a cohort of SN patients with mixed etiologies. It is not yet clear whether there is a correlation between etiology and extent of cord atrophy. To address this issue, quantitative studies with a single-etiology group of patients would be important. Recently, Chevis et al21 performed such a study in a large cohort of Friedreich’s ataxia patients, which is the prototypical genetic SN. Authors demonstrated cord atrophy and flattening as well as a direct correlation between cord area reduction and clinical disability. Disease duration also correlated with cord atrophy in these patients. Cerebellum
Cerebellar atrophy has been considered a rare finding in SN patients, but it is possibly overlooked. Damasceno et al22 used a semiautomated method to measure volumes of the whole cerebellum (CEV) as well as the anterior lobe of the cerebellum (ALV). These values were then normalized to whole brain volume and compared between healthy controls and patients. Authors failed to identify group differences, but some patients had extremely reduced volumes (smaller than 2.5 SDs of the control group mean values). This suggests that cerebellar atrophy is not common, but may be found in occasional SN patients (Fig 4).
Dorsal Root Ganglion
DRG are very small structures, so that most MRI acquisition protocols fail to image them properly. Recently Bao et al15 were able to visualize DRG using MEDIC images. They then modeled DRG at C7 level as an ellipse and showed that its normalized axes (“length” and “width” of the DRG) were reduced in SN patients, when compared to healthy control and disease control groups. Although promising, these results need to be replicated before widespread use.
Other Findings Some MRI-based studies performed in the early stages of SN demonstrated alterations outside the dorsal columns. T2 weighted signal changes were observed in a patient with Sjogren’s syndrome on the medial lemniscus and also in the ¨ pons and cerebellar peduncles with a normal cervical image, reflecting proximal sensory pathways impairment and a more diffuse immune-mediated attack.23 Bao et al15 demonstrated high signal intensity in cervical nerve roots without changes in dorsal columns in a patient with recent onset of SN, and in a follow-up exam 7 months after SN onset, showed hs-T2WI in dorsal columns and decreased high signal intensity. Bao et al15 also attempted to quantify DRG signal intensity at C7 on MEDIC images. They found significantly higher intensity in SN patients compared to both healthy and disease controls. In contrast, when using TIRM images (turbo inversion
Casseb et al: Neuroimaging in Sensory Neuronopathy
707
recovery magnitude sequence), no difference was found in signal intensity from C7 nerve roots, but the diameter was smaller in SN patient group.
Final Remarks Spinal cord MRI should not be considered in isolation to assist in the diagnosis of SN. One must take into account clinical and neurophysiological data. However, there are challenging patients who may benefit from detailed MRI analysis, such as those presenting purely sensory abnormalities but with a worsening gradient toward the distal legs. MRI is widely available and noninvasive procedure in contrast to the techniques that rely on biopsy analysis. We thus believe that MRI should be considered as an additional tool in the work-up of SN patients.
Future Perspectives We somehow understand basic hs-T2WI characteristics, but some of them are still unclear. Future work should address specific questions, such as the time of hs-T2WI onset and whether there is any etiology-specific pattern of abnormalities. To accomplish that, longitudinal investigations would be necessary, and it would require assessment of patients in the early stages of the disease (when there is no hs-T2WI) until the hs-T2WI is detected. The major issue is to find patients at early stages, since they are normally brought into medical attention in advanced stage, making it difficult to perform such studies. Neuroimaging progress is uncovering new fields to be explored. MR diffusion tensor imaging (DTI), for instance, is being employed in the investigation of other spinal diseases. Some interesting results were reported for amyotrophic lateral sclerosis showing reduction of fractional anisotropy and increased ADC values in the lateral columns of the cervical spinal cord.24 Since SN also involves degeneration of the long spinal tracts, we hypothesize that cord DTI might also prove useful to study SN patients. Moreover, higher field MR equipment may also favor investigation, since image quality (spatial resolution and signalto-noise ratio) is deeply impacted. In line with that, a recent study with an ALS patient, using a 7T MR system, showed impressive images of corticospinal tract damage at cervical spinal cord.25 Most of the aforementioned papers did not focus on brain MRI abnormalities. Nonetheless, cerebellar damage has been already reported in a few patients. We believe that encephalic involvement in SN has been overlooked and therefore, this is certainly an area to be further explored. Functional magnetic resonance imaging (fMRI) is another promising technique to investigate neuronal behavior both during mental rest and task conditions. Weeks et al26 investigated 3 SN patients with absent proprioception using fMRI and PET scans, and verified that during finger movements, patients did show cerebellar activation both in fMRI and PET examinations, even with absent proprioception sense. This favored the hypothesis that cerebellum coordinates voluntary movement through the integration of multiple inputs and not only the sensory ones. Overall, this study emphasizes that SN is a useful human model to study sensory-motor integration through MRI-based techniques. In the long term, such studies might be able to identify relevant targets in the brain amenable
708
for therapeutic interventions (eg, neuromodulation) in SN patients. The authors thank CAPES and FAPESP for funding this project, and Abril Jimenez for the language revision.
References 1. Denny-Brown D. Primary sensory neuropathy with muscular changes associated with carcinoma. J Neurol Neurosurg Psychiatry 1948;11:73-87. 2. Martinez AR, Nunes MB, Nucci A, et al. Sensory neuronopathy and autoimmune diseases. Autoimmune Dis 2012;2012:873587. 3. Sghirlanzoni A, Pareyson D, Lauria G. Sensory neuron diseases. Lancet Neurol 2005;4:349-61. 4. Colli BO, Carlotti CG Jr, Assirati JA Jr, et al. Dorsal root ganglionectomy for the diagnosis of sensory neuropathies. Surgical technique and results. Surg Neurol 2008;69:266-73. 5. Garcia RU, Ricardo JA, Horta CA, et al. Ulnar sensory-motor amplitude ratio: a new tool to differentiate ganglionopathy from polyneuropathy. Arq Neuropsiquiatr 2013;71:465-9. 6. Lauria G, Pareyson D, Sghirlanzoni A. Neurophysiological diagnosis of acquired sensory ganglionopathies. Eur Neurol 2003;50:14652. 7. Camdessanche JP, Jousserand G, Ferraud K, et al. The pattern and diagnostic criteria of sensory neuronopathy: a case-control study. Brain 2009;132:1723-33. 8. Okumura R, Asato R, Shimada T, et al. Degeneration of the posterior columns of the spinal cord: postmortem MRI and histopathology. J Comput Assist Tomogr 1992;16:865-7. 9. Sobue G, Yasuda T, Kumazawa K, et al. MRI demonstrates dorsal column involvement of the spinal cord in Sjogren’s syndromeassociated neuropathy. Neurology 1995;45:592-3. 10. Waragai M, Takaya Y, Hayashi M. Serial MR findings of chronic idiopathic ataxic neuropathy. J Neurol Sci 1997;151:93-5. 11. Lauria G, Pareyson D, Grisoli M, et al. Clinical and magnetic resonance imaging findings in chronic sensory ganglionopathies. Ann Neurol 2000;47:104-9. 12. Mori K, Koike H, Misu K, et al. Spinal cord magnetic resonance imaging demonstrates sensory neuronal involvement and clinical severity in neuronopathy associated with Sjogren’s syndrome. J Neurol Neurosurg Psychiatry 2001;71:488-92. 13. Damasceno A, Franca MC Jr, Nucci A. Chronic acquired sensory neuron diseases. Eur J Neurol 2008;15:1400-5. 14. Franca MC Jr, D’Abreu A, Zanardi VA, et al. MRI shows dorsal lesions and spinal cord atrophy in chronic sensory neuronopathies. J Neuroimaging 2008;18:168-72. 15. Bao YF, Tang WJ, Zhu DQ, et al. Sensory neuronopathy involves the spinal cord and brachial plexus: a quantitative study employing multiple-echo data image combination (MEDIC) and turbo inversion recovery magnitude (TIRM). Neuroradiology 2013;55:41-8. 16. Mori K, Iijima M, Koike H, et al. The wide spectrum of clinical manifestations in Sjogren’s syndrome-associated neuropathy. Brain 2005;128:2518-34. 17. Sheikh SI, Amato AA. The dorsal root ganglion under attack: the acquired sensory ganglionopathies. Pract Neurol 2010;10:326-34. 18. Gorson KC, Herrmann DN, Thiagarajan R, et al. Non-length dependent small fibre neuropathy/ganglionopathy. J Neurol Neurosurg Psychiatry 2008;79:163-9. 19. Chong J, Di Rocco A, Tagliati M, et al. MR findings in AIDSassociated myelopathy. AJNR Am J Neuroradiol 1999;20:1412-6. 20. Hemmer B, Glocker FX, Schumacher M, et al. Subacute combined degeneration: clinical, electrophysiological, and magnetic resonance imaging findings. J Neurol Neurosurg Psychiatry 1998;65:822-7. 21. Chevis CF, da Silva CB, D’Abreu A, et al. Spinal cord atrophy correlates with disability in Friedreich’s ataxia. Cerebellum 2013;12:43-7.
Journal of Neuroimaging Vol 25 No 5 September/October 2015
22. Damasceno A, Franca MC Jr, Cendes F, et al. Cerebellar atrophy is infrequently [corrected] associated with non-paraneoplastic sensory neuronopathy. Arq Neuropsiquiatr 2011;69:602-6. 23. Damasceno A, Franca MC Jr, Zanardi VA, et al. Brainstem involvement in Sjogren’s syndrome-related sensory neuronopathy. J Neuroimaging 2010;20:397-9. 24. Wang Y, Liu L, Ma L, et al. Preliminary study on cervical spinal cord in patients with amyotrophic lateral sclerosis using MR diffusion tensor imaging. Acad Radiol 2014;21:590-6.
25. Cohen-Adad J, Zhao W, Keil B, et al. 7-T MRI of the spinal cord can detect lateral corticospinal tract abnormality in amyotrophic lateral sclerosis. Muscle Nerve 2013;47:760-2. 26. Weeks RA, Gerloff C, Honda M, et al. Movement-related cerebellar activation in the absence of sensory input. J Neurophysiol 1999;82:484-8. 27. Takahashi Y, Takata T, Hoshino M, et al. Benefit of IVIG for longstanding ataxic sensory neuronopathy with Sjogren’s syndrome. IV immunoglobulin. Neurology 2003;60:503-5.
Casseb et al: Neuroimaging in Sensory Neuronopathy
709
