MRI T2-Hyperintense Signal Structures in the Cervical Spinal Cord: Anterior Median Fissure versus Central Canal in Chiari and Control—An Exploratory Pilot Analysis

DISCUSSION

Patients with Chiari I are known to have a high incidence of hydrosyringomyelia. Therefore, identification of round foci on axial images (HIF) and thin craniocaudad channels on sagittal images (SL), both usually considered manifestations of the central canal, is anticipated. However, the higher incidence of identification and frequency of an AMF in patients with Chiari I compared with controls has not been previously reported. AMF/pt. increased progressively from all (2.7/pt.), to HIF+AMF (4.3/pt.), to AMF>HIF (4.7/pt.), and to SL AMF>HIF in patients with Chiari I (5.1/pt.). The ratio of HIF/AMF decreased among patients with SL, while AMF/pt. and AMF>HIF/pt. were increasing. AMF>HIF/pt. was highest among patients with an SL (2.9/pt.). Increasing AMF/pt., a decreasing HIF/AMF ratio, and increased AMF>HIF/pt. in patients with SL represent a stronger association of AMF-to-SL identification than suggested in the original non-Chiari I study,¹ in which the AMF>HIF/pt. were not counted.

It may be argued that with increases in AMF, identification of an AMF>HIF becomes more likely. However, the observed AMF>HIF/pt. was 47% greater in number than expected on the basis of HIF and AMF numbers for all patients with Chiari I and 58% greater for patients with an SL with AMF>HIF. The correlation of increasing AMF to AMF>HIF/pt. numbers in the setting of a decreasing HIF/AMF ratio supports an etiologic relationship between the visible cleft of an AMF and the round HIF of a potential channel or SL and indicates a potential anatomic connection of the 3 structures. This raises the question of whether some thin craniocaudad channels simulating the central canal on midline sagittal imaging are actually due to widened bases of AMFs. While numbers of AMF>HIF/pt. were not counted in the original non-Chiari I study,¹ other measures of HIF, AMF, and SL occurrence in that study are generally consonant with current control findings.

Despite an inability to resolve and separate without question the base of the AMF and HIF or the central canal individually on the same axial image in routine clinical scanning,⁷ the relationships in our previously published analysis of HIF, AMF, and thin SL in a non-Chiari I MR image population provided not only a baseline control for their incidence of identification but also a confirmation of a relationship among AMF, HIF, and SL.¹ This confirmation led us to believe that depictions of HIF and AMF are not totally independent occurrences of 2 different structures (the AMF and the central canal) but rather identification of different manifestations of the same structure: the sagittal cleft of the AMF as well as its wider base, in some instances.

In our prior review of 35 patients without Chiari I with both cervical MR imaging and cervical CT myelography (CTM), iodinated contrast media–filled AMFs were more common than HIF on CTM.¹ When present, HIF on CTM appeared almost exclusively as a focal dilation of the base of an AMF on CTM. A number of AMFs on CTM were seen at the same spinal level where MR imaging HIF were identified when an MR imaging AMF had not been identified. In several instances, the AMF on CTM was wide, possibly freely communicating on multiple, contiguous, consecutive axial images, while seen as an atypical broad, indistinct channel on a sagittal image due to partial volume effects on MR imaging. These CTM data are also consistent with HIF representing a wide base of the AMF.^8,9

Our findings raise the question of the physiologic origin and basis for imaging findings of the increased HIF, AMF-related channel, or SL simulating hydromyelia on MR imaging. CSF flow dynamics are known to be altered in patients with Chiari I, resulting in imaging of complex, pulsatile CSF motion that is subject to greater or lesser bright-T2 conspicuity with certain pulse sequences, even varying between systole and diastole.^10⇓-12 Pulsatile CSF flow effects directed in a Craniodaudal or transverse radial fashion that have been suggested to cause syringomelia¹³ may be transmitted to the AMF, whose entire cervicomedullary course may not be subject to equal directions and magnitude of force. This possibility may be suggested by variable signal changes of flow and flow-related signal loss on axial images (Figs 3 and 4). Pial/arachnoidal coverings and vessels within the AMF¹⁴ may limit communication with the subarachnoid space and secondarily alter local transverse pressure transmission, but they may not limit and may even amplify craniocaudad effects (Figs 1–4). The effects may vary among subjects, and their depiction may vary among field strengths and imaging parameters applied. A formula for the production and/or loss of T2-signal in the AMF has yet to be elucidated.

The main limitation of this study lies in subjective observations of a single, potentially biased reader. That the current explorative study identified control SL incidence similar to prior non-Chiari study (28% versus 25%) as well as similar AMFs and HIF/pt. (3.9 versus 3.7 HIF, and 2.7 versus 2.3 AMF/pt.) suggests that scanners and reader performed similarly in this regard. However, the higher incidence of both HIF and AMF in 25 controls was seen compared with the original 6-scanner retrospective control study with 2 readers (96% versus 66% HIF, 88% versus 60% AMF). Scanner and reader performance may explain this discrepancy. Wide ranges of contributions to the identified HIF and AMF totals were found in the original 2-reader, 6-scanner analysis of controls without Chiari I: 5.9%–15.4% for HIF and 5.9%–16.7% for AMF among the 6 individual scanners (Online Supplemental Data).¹ At these rates of identification, a hypothetic similar top-performing scanner might record nearly 100% HIF and/or AMF (16% × 6 scanners = 96%) in a Chiari I population, and 6 similar lowest-performing scanners, just 36% (6% × 6 = 36%). The single scanner of the current study did perform at a similar level for HIF and AMF incidence compared with the highest-performing scanner in the prior 6-scanner non-Chiari I study. However, no single scanner performed best for identification of HIF, AMF, and SL, and no single scanner performed worst for individual structures in the prior study.

That the percentage of SLs was similar for both control subgroups without Chiari I is also consistent with the anticipated range of scanner performance (identified as 12.7%–34.2% in the 6-scanner study) and with the κ analysis in the original study, in which both readers agreed that <70% of patients did not have an SL. The small numeric increase of SLs in patients with Chiari I without operations versus controls is consistent with the original study, in which scanners that identified high numbers of HIF and AMFs did not identify high numbers of SLs. The discrepancy may reflect case selection, in which many patients with Chiari I with larger channels were already excluded from analysis. That a higher percentage of SLs was identified in postoperative patients with Chiari I may indicate a bias in favor of an operation for patients with hydrosyringomyelia preoperatively. The numeric increase of AMF and AMF>HIF in patients with Chiari with and without an operation remains of interest. A more comprehensive review of a larger population of pre- and postoperative scans in patients with Chiari I is required to answer these questions.

The observations reported here may appear to be of little immediate and direct clinical importance, because thin channels of <3 mm in patients with Chiari I are thought to be of little clinical consequence.^5,15 However, MR imaging reports attributing an SL or thin channel in a patient with Chiari I to hydro- or syringomyelia may increase patient concern, causing him or her to consider any neurologic symptom as potentially related, leading to unnecessary restudy. To solidify a relationship between conspicuous AMFs and such SLs or channels would serve to diminish concerns based on historical hydrosyringomyelic relationships.