An Optimized CT Protocol for Detecting Suspected Cauda Equina Syndrome: A Comparative Analysis with MRI

DISCUSSION

This study prospectively evaluated the diagnostic accuracy of an optimized CT lumbar spine protocol in identifying severe lumbar disc herniation in patients suspected of having acute severe neural compression. To our knowledge, this is the first study prospectively comparing CT with MR imaging for the evaluation of suspected disc herniation. With minor modifications to a routine lumbar spine protocol, CT performed well, with high sensitivity and specificity (97% each) using MR imaging as the criterion standard.

Notably, 37% (n = 22) of the cohort demonstrated no significant stenosis, underscoring the clinical challenge of accurately diagnosing suspected CES and the pivotal role of imaging in its evaluation. In a further 37% (n = 22) of patients, conservative management was sufficient, indicating that not all cases of suspected disc pathology necessitate emergency surgical intervention even if the patient presents to the ED with acute severe symptoms. In scenarios in which CT excludes the presence of neural compression, these findings suggest that patients can be safely directed to outpatient or in-hours MR imaging for further evaluation. Conversely, when CT indicates probable compression, it can effectively guide the need for emergent MR imaging, optimizing patient triage in emergency settings when MR imaging availability is limited.

Lumbar disc herniation is a leading cause of CES, a condition necessitating urgent surgical intervention.² The clinical diagnosis of CES is notoriously challenging, marred by its complex presentation, potential overlap with other lumbar pathologies, and confounding factors such as opioid-induced urinary retention. Despite an effort to pinpoint symptom combinations indicative of compressive lesions, clinical features alone have proved to be unreliable predictors.^11⇓⇓-14 A systematic review revealed that the prevalence of CES based on positive MR imaging findings following clinical evaluation varies widely from 14% to 48%.¹⁵ This variability, coupled with the generally low level of evidence in the CES literature, which has been echoed by other authors,¹⁶ shows that clinical assessment alone is insufficient, necessitating reliable imaging for definitive diagnosis.

The limited availability of MR imaging, especially in emergency settings, compounds the challenge of managing potential CES cases. This limitation is exemplified by the dilemma of “scan-negative” patients, in whom clinical suspicion of neural compromise exists, but MR imaging fails to confirm a compressive lesion.¹⁷ Our study underscores the value of CT in this context. With its widespread availability and rapid imaging capability, CT can significantly enhance the efficiency of emergency care. For instance, a UK study reported that 24% of out-of-hours spinal center referrals required transfers for MR imaging, incurring substantial costs estimated at £6000 (US $7687.20) per transfer.¹⁸ Notably only 2.9% of patients (n = 45/1529) in this study required surgical decompression. Moreover, in a cohort of 250 neurosurgical referrals, evaluation by MR imaging was delayed until the following day for 82% of out-of-hours cases (n = 60/73).¹⁹ Delays such as these, while often unavoidable under current resource limitations, contradict recommended clinical pathways of care, which demand urgent access to imaging. For example, the national suspected cauda equina syndrome pathway published by the Getting It Right First Time Group in the United Kingdom recommends that patients presenting with suspected CES should undergo MR imaging within 4 hours.²⁰ Given the current limitations in MR imaging availability and the ubiquity of CT, incorporating our CT protocol could improve care efficiency, especially in urgent scenarios. This integration would allow a more pragmatic approach to managing suspected cases of CES, aligning with established clinical pathways while awaiting more widespread MR imaging accessibility.

Several studies have explored the comparative effectiveness of CT and MR imaging in assessing lumbar and foraminal stenoses. Our findings align with those of Peacock and Timpone,⁶ who reported that CT identified CES with a sensitivity of 98%, specificity of 86%, positive predictive value of 72%, and a negative predictive value of 99%. The notably higher specificity and positive predictive value in this study may be, in part, due to differences in methodology, with the current study being prospective in nature, using a CT protocol optimized for lumbar disc assessment and the CT and MR imaging being performed during the same ED episode, compared with being obtained within 48 hours of each other as in the retrospective study by Peacock and Timpone. In addition, the smaller sample size in our study may be contributory (59 versus 151). The high sensitivity and negative predictive value observed in our study reinforce the utility of CT, not as a replacement for MR imaging but as a valuable adjunct in accurately and safely triaging patients for emergency MR imaging.

Alternative imaging modalities have also been investigated to attempt to improve the triage of individuals with suspected CES for urgent MR imaging. A prospective study of 260 cases evaluated bladder ultrasound (to assess a postvoid residual volume of > 200 mL) as an adjunct to clinical assessment for suspected CES. The authors found that the addition of bladder scanning improved diagnostic accuracy for predicting radiologic CES (sensitivity, specificity, and positive and negative predictive values of 94.1%, 66.8%, 29.9%, and 98.7%) compared with the presence of red flag symptoms such as bilateral sciatica alone.²¹ While ultrasound also has the benefit of relative ubiquity and is not associated with radiation exposure, CTSS overall demonstrates superior diagnostic accuracy with markedly improved specificity and positive predictive values and similar sensitivity and negative predictive values.

Despite the superior soft-tissue contrast of MR imaging, previous studies have also demonstrated the efficacy of CT in assessing canal stenosis and cord compression. Alsaleh et al²² highlighted the excellent intraobserver agreement of CT (κ = 0.96) in detecting canal stenosis. Similarly, for foraminal stenosis, Kang et al²³ found that the intra- and interobserver agreement of CT paralleled that of MR imaging, especially among experienced readers.

In assessing canal stenosis, expert musculoskeletal radiologists commonly rely on qualitative assessments, as highlighted in the Delphi meeting findings.⁹ This approach was adopted in our study, aligning with prevalent clinical practice. Nonetheless, the challenge remains in defining an optimal measurement for spinal stenosis, given the variety of available methods.^9,10 We incorporated cross-sectional area measurements of <1 cm² as an objective adjunct to our primary qualitative analysis, in line with established quantitative criteria for spinal stenosis. This decision is supported by evidence suggesting that smaller canal cross-sectional areas correlate with more severe symptoms.²⁴ Our analysis revealed that CT measurements were generally consistent with those of MR imaging, particularly in stenotic areas, as detailed in Tables 2 and 3. Subgroup analysis further elucidated the accuracy of CT, especially in cases with cross-sectional areas of <1 cm², in which measurement discrepancies were smaller. This finding bolsters confidence in the reliability of CT for diagnosing canal stenosis and cord compression. Additionally, thecal sac effacement of >50% was used as a validation of the findings of Peacock and Timpone,⁶ concurring with the MR imaging findings in 99%. However, for foraminal stenosis, the agreement of CT with MR imaging was slightly lower at 75%, attributable to the broader grading scale used for foraminal assessment. Notably, in instances of severe foraminal stenosis (grade 3), CT findings aligned with MR imaging findings in 90% of cases, underscoring its efficacy in detecting clinically significant compressions.

Identifying the optimal approach for assessing canal stenosis remains an area of ongoing research. Recent advancements include novel techniques that use the ratio of disc-to-canal cross-sectional areas and the anterior-to-posterior diameter, offering promising results in stenosis identification and its correlation with clinical symptoms.²⁵ Ratio-based metrics such as these that are normalized against anatomic and demographic differences may allow more robust assessment of stenosis in future studies.

A notable consideration is the radiation exposure associated with CT, which is a drawback compared with MR imaging. However, the use of our CT protocol is strategically limited to cases of suspected spinal emergencies when MR imaging is not immediately accessible, a common scenario in many regions, including ours. The mean DLP of CTSS was found to be 579 mGy × cm. This was similar to our local routine lumbar spine CT dose and is comparable with and, in some cases, lower than the CT lumbar spine DLPs reported in the literature. For example, the national dose reference level for CT of the lumbar spine in Australia for 2020 was 670 mGy × cm.²⁶ In a more recent study from Jordan, the reported 75th percentile for lumbar spine CT was 967.7 mGy × cm,²⁷ which is considerably higher than that in this study (602 mGy × cm). In general in the adult population, the dose associated with a CT lumbar spine is considered acceptable for the evaluation of suspected vertebral fractures. Given the potential severity of the pathology in suspected cauda equina syndrome and the resource limitations associated with emergency MR imaging, the equivalent exposure is justifiable provided the diagnostic test is accurate. Advances in dose-reduction techniques, such as photon-counting and artificial intelligence reconstruction, are anticipated to mitigate this concern, achieving comparable image quality at lower radiation doses. Our department plans to explore these innovations as they become available.

Regarding limitations, our study, while promising, had a relatively small patient sample size. Conducting larger, multicenter studies would provide more comprehensive data to validate our findings. The sample size, in terms of vertebral levels and neural foramina assessed, was substantial, yet predominantly negative for stenosis. This feature could impact the generalizability of our results, but it is noteworthy that our sample reflects the typical ED population with suspected acute nerve root compression, lending relevance to daily clinical practice. We also recognize that the use of an optimized CT protocol might create concerns about generalizability. Nevertheless, the modifications we introduced are feasible with current CT technology and can be implemented on most modern scanners. This approach, while specific to our study, adheres to principles and parameters that are achievable in various clinical settings, potentially broadening its utility in similar diagnostic contexts.

The absence of a universally accepted method for assessing and grading canal stenosis poses another challenge; to address this issue, we used multiple assessment techniques, including central canal stenosis >50%, foraminal stenosis, cross-sectional area, and qualitative assessment, enhancing the credibility of our CT evaluations. While the reader was blinded to the clinical information and outcome, the knowledge that the imaging was being performed as part of an ED assessment for acute radicular symptoms and suspected CES could introduce bias toward calling disc pathology, though the later correlation of pathology with management that demonstrated that all lesions requiring emergent decompression were correctly identified supports the radiologist’s assessment. While our study demonstrated the utility of CT in identifying cases of nerve root compression predominantly caused by disc herniation requiring urgent surgical intervention, it is important to recognize its limitations. Specifically, our study did not assess the diagnostic accuracy of CT for other causes of CES, such as intradural neoplasms, infection, and inflammation. Consequently, the generalizability of our findings to these conditions may be limited. This study does not propose CT as a substitute for MR imaging across all spinal pathologies but highlights its potential in urgent cases in which MR imaging accessibility is constrained. The inclusion of multiple raters, with varied experience levels, in future studies would enhance the robustness and generalizability of our findings. The single-reader model of this study presents a limitation in terms of variability and broader applicability.