Magnetic resonance imaging and outcome.

The optimal management of cervical spondylotic myelopathy (CSM) continues to be a subject of considerable controversy. One of hottest areas of debate is focused on the timing of surgical intervention based on the clinical presentation of CSM. Currently available data would suggest that surgery is clearly indicated in patients with moderate or severe CSM, whereas there is greater uncertainty in patients with mild CSM or those with minimal functional impairment.3 Given this uncertainty in the optimal management of patients with CSM, there has been great interest in the potential use of MR imaging to predict which patients could benefit the most from surgical intervention, and to identify factors that could predict long-term outcomes in this disease. Magnetic resonance imaging is a key preoperative assessment tool in the setting of CSM. A comprehensive knowledge of the importance of the changes associated with this imaging modality is essential for patient selection and education in regard to postoperative recovery expectations. There is emerging evidence that suggests that the presence of T1 and T2 signal, segmentation of T2 signal, and the presence and severity of spinal cord compression all have poor prognostic implications.1,4,6,7 The literature, however, has inherent flaws in the lack of uniformity of patients with CSM, variable inclusion and exclusion criteria, and insufficient and blunt tools for its assessment. The study by Yagi et al.9 in this issue of the Journal of Neurosurgery: Spine attempts to shed further light on the prognostic significance of intramedullary signal changes on baseline MR imaging studies. In this study, the authors show that the presence of T2 signal change is associated with significantly worse clinical outcome as measured by the Japanese Orthopaedic Association (JOA) score and recovery ratio. In particular, the upper-extremity parameters were shown to be significantly worse in the setting of increased T2 signal change. Furthermore, using regression analysis, the authors found that instability of the cervical spine (as measured by dynamic radiographs) and ventral compression were important independent risk factors for worsened clinical outcomes. Moreover, in patients in whom the T2 signal worsened postoperatively (17 of 50), a worse outcome was reported. A key take-home message from the paper by Yagi et al.9 is that when treating CSM, it is critical to address any preexisting cervical instability as well as the offending compressive pathological entity. However, other points in the paper also require commentary. First, in our view, a focal ventral pathological entity is best treated by a direct anterior approach (discectomy/corpectomy and reconstruction) as opposed to a posterior cervical laminoplasty. Second, the subcategorization of T2 signal change, either in terms of area or shape, is not an easy undertaking. In the paper by Yagi et al., one radiologist graded the MR images, and no assessment of interobserver agreement was undertaken. The reported subclassification of intramedullary MR imaging signal changes into “snake-eye” and “pencil shape” has not been validated. Moreover, the 3 categories describing the longitudinal extent and associations of T2 signal change (that is, limited to 1 disc, beyond 1 disc, and involvement of T1 low signal intensity) are somewhat arbitrary. Indeed, the authors’ final conclusion that “signal changes on T1as well as T2-weighted MR sequences obtained in patients with [cervical compression myelopathy] indicated a poor long-term prognosis” is not an entirely valid one, based on the results presented. The patients with T1 low signal intensity in this study showed no statistically significant difference in mean modified JOA (mJOA) scores and recovery ratio (p = 0.069 and 0.079, respectively). Third, the increase in the size of the T2 signal change was neither qualified nor quantified using image analysis software. As we have found in our own work, the measurements of diffuse T2 signal area, particularly adjacent to an area with a compressive pathological entity and bright CSF T2 signal, is not trivial. The key point here is that the diffuse and bright T2 signal changes may represent different pathophysiological mechanisms or a single process in evolution. The functional recovery of the patients in this study was assessed by means of JOA score and recovery ratios. Although the JOA scale and its modified counterpart (the mJOA2) is currently the so-called gold standard, there is emerging evidence that patients with CSM lose and J Neurosurg Spine 12:56–58, 2010 See the corresponding article in this issue, pp 59–65.