Prognostic Value of Age and Early Magnetic Resonance Imaging in Patients with Cervical Subaxial Spinal Cord Injuries
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0028-3886.304104
Source of Support: None, Conflict of Interest: None
Keywords: MRI, radiology, recovery, rehabilitation, spinal cord injury
Patient age represents an important predictive factor for outcome in several neurological conditions, although its role in acute traumatic spinal cord injury (SCI) is still controversial.,, Confirming the discriminative capacity of patient age as a predictor for neurological and functional outcomes in SCI, however, would have important consequences for national policies and international guidelines.
The extension of intramedullary contusions on initial magnetic resonance imaging (MRI) has been identified as a robust predictive factor in cases of traumatic cervical SCI.,,,,, Nonetheless, one question remains to be addressed: can the extension of the intramedullary lesion be considered an independent predictor of outcome, or does age represent a confounding factor? A comparison of the extension of SCI injury across different age groups merits further attention, as it perhaps would provide additional insight on how to optimize clinical and surgical management of SCI patients.
We designed the present study with a twofold objective: a) to assess the prognostic role of patient's age versus radiological features, and b) to determine clinical and radiological thresholds for outcome prognostication in a cervical SCI population.
A retrospective analysis of patients with cervical SCI treated between January 2008 and December 2017 in three European tertiary trauma centers was performed. The study was conducted in compliance with ethical guidelines and applicable patient privacy regulations. The work was presented to the investigational review board of all institutions, which waived the need for informed consent in terms of the retrospective design of the study and the application of strict patient privacy regulations operating in our centers.
Patients older than 15 years with at least 12 months follow-up were considered for study inclusion. Inclusion criteria included: a computed tomography (CT) and MRI as part of the initial investigations (performed before any surgical intervention, and in every case within 48 h after the injury), regardless of the mechanisms of injury, initial American Spinal Injury Association (ASIA) Impairment Scale (AIS). Given the potential burden caused by other traumatisms, patients with severe traumatic brain injury, upper cervical (C1–C2), thoracic, and lumbar SCI were excluded.
Clinical assessment at admission and follow-up
Demographic data, mechanism of injury, and initial neurological status were extracted from handwritten clinical notes and electronic charts. Clinical information obtained from the initial neurological assessment and at the one-year follow up were reported and analyzed according to the AIS grade. “Poor neurological outcome” was defined as a complete SCI (AIS grade A) at 1-year-follow-up. For assessing the functional status, we chose the Walking Index for Spinal Cord Injury II (WISCI II) for its superiority to test gait and walking recovery. A score inferior to 1 in the WISCI II was considered as “poor functional outcome”. Previous results suggest that the WISCI II should be a very useful outcome measure for detecting changes in walking function following chronic SCI.Since the objective was to evaluate functional capacity at the long term, the authors choose the WISCI II, since this scale was initially designed to assess the ability to walk-in patients with chronic SCI.
Every center relied on helical CT and a 1.5 Tesla MRI for the initial neuroradiological investigation. Since some radiological MRI findings, such as length of edema, are dependent on the time of MRI acquisition after trauma, most of the previous studies have recommended to perform the MRI within the acute phase (48 h).,, As such, a threshold for the timing of MRI acquisition was established at 48 h.,
All MRIs were acquired with a slice thickness of 3 mm for axial and sagittal planes. The investigational protocol at each institution included axial and sagittal T1- and T2-weighted images, axial gradient echo; and sagittal short tau inversion recovery (STIR) sequences.
The length of the intramedullary lesion was defined as the distance, in millimeters, between the most cephalic and the most caudal extent of the cord signal hyperintensity,,,[Figure 1].
Integrity or disruption of soft tissues, including the anterior and posterior longitudinal ligament, ligamentum flavum, and intervertebral discs (ALL, PLL, LF, and ID, respectively), were assessed by analyzing T2-weighted imaging and STIR sequences in a sagittal plane. Soft tissues were classified as “normal” or “injured”. The injury was defined as complete discontinuity of the MRI signal in all appropriate imaging cuts.
The assessment of MRI was consensually made by an experienced spine neurosurgeon and experienced neuroradiologist. Both observers were blinded to the patients' clinical and epidemiological data.
Clinical and surgical management
Patients were managed in accordance with the Advanced Trauma Life Support (ATLS) protocol. The administration of steroids according to the NASCIS II protocol was consistent across patients. The indication for surgery and the choice of the surgical approach were made by the attending neurosurgeon on call. Informed consent was obtained from the patient and/or the next of kin or caregiver.
Data were analyzed using nonparametric tests. To determine the relationship between the patients' age and the severity of the cervical trauma, the quantitative variable “age” was compared with imaging quantitative variables such as “length of the intramedullary lesion” and “number of soft tissues affected” and clinical variables (AIS at admission and at follow-up). The number of soft tissue structures injured (ALL, PLL, LF, and ID) ranged from 0 to 4. AIS was considered a quantitative variable. Since the normal distribution of the sample cannot be ensured, the Spearman rho test was used to evaluate the relationship between “age” and “the length of intramedullary lesion”, “AIS grade at admission”, and “number of soft tissues affected'.
To detect if the patient's age was a predictive factor of the neurological function independent of the length of the intramedullary lesion, patients were classified into two groups according to the length of intramedullary lesion seen on the initial MRI. The threshold to dichotomize patients according to the length of the intramedullary lesion was set at 36 mm as this value was identified as a meaningful predictor of prognosis in a previous study8. Spearman's rho was used to compare the correlation between the patient's “age” and “length of the lesion”, “the number of soft tissues injured”, and the AIS at admission and at follow-up in both groups of lesion length (>36 mm and <36 mm).
The Mann–Whitney test was used to detect if there was any difference in mean age in patients with complete spinal cord injury and in those that were unable to walk or stand (Walking Index for Spinal Cord Injury II –WISCI II- <1) at follow up. Graphs were drawn to determine whether there was an appropriate cutoff value related to worse neurological (complete SCI) and functional prognosis (unable to walk or stand, WISCI II <1) for the variable “age”. For each possible cut point of the measure under investigation (age), there is an associated sensitivity and specificity. These can be summarized using a ROC curve [Figure 2], in which each point of the curve represents a different cut point of the measure. The optimal cut-point was obtained by means of determining the maximum Youden index, which is the point in which sensitivity and specificity are maximal and therefore is the point in the ROC curve with the largest vertical distance from the line of chance.
To detect if there were any differences in functional outcome (WISCI II <1 at 1 year-follow-up) regarding the treatment received comparisons using Fisher exact test were done.
The threshold for statistical significance was P < 0.05, and calculations were done using SPSS v22 (IBM, etc., Armonk, NY, USA) and Medcalc v17 (Medcalc Software, Ostend, Belgium).
Two-hundred-twenty-eight patients were retrospectively reviewed. After applying exclusion criteria, the final analysis included 134 patients (mean age 43.4 years, range 17–87 years, 80% males).
High-speed road traffic accidents (RTA) were the mechanism of injury in 72 patients (52% of the population), whereas 45 (33%) patients had cervical SCI as a result of mechanical falls. The mean age in patients involved in high-speed RTA was 42 years (range 17–71), whereas it was 58 years (range 36–87 years) in those suffering a mechanical fall.
Seventy patients were managed conservatively and 64 surgically. The incidence of polytrauma was 20 and 50 percent in the conservative and surgical population (P < 0.001); with a predominance in younger patients (mean age for patients with associated polytrauma vs without polytrauma 34 vs 49 years, P < 0.001). No differences in functional outcome (WISCI II < 1) were seen between patients treated surgically and conservatively (P = 0.13), nor for different surgical approaches (P = 0.29).
The patient's neurological status by means of the AIS grade at admission and follow up are represented in [Table 1]. Statistical correlations between age and other clinical and imaging variables in the overall population are presented in [Table 2]. Young age was correlated with larger intramedullary lesions, although this relationship did not reach statistical significance (P = 0.22). There was no significant correlation between patient age and neurological function neither at the initial exam nor at follow-up. Similar results were obtained for the group of patients with lesions larger than 36 mm [Table 3]. In the group of patients with lesions lesser than 36 mm [Table 4], patients with a more advanced age correlated with worse neurological exams on the AIS impairment scale at admission (P = 0.006). Additionally, in this group (intramedullary lesions smaller than 36 mm) age of patients with complete SCI at 1-year follow-up was significantly older than for those patients with incomplete SCI (63 vs 45.6-year-old; P = 0.04) [Table 5]. Similarly, patients with the poor functional outcome (WISCI II < 1) were older than those with a good functional outcome (64 vs 44-year-old; P < 0.001).
The discriminative capacity of the patient's age for a poor neurological (ASIA A) and functional outcome (WISCIS II <1) at follow-up are summarized in [Figure 2] by means of the area under the ROC curve. The ROC curves determined excellent discriminative capacity for a poor functional capacity [AUC = 0.86;(95% CI 0.780–0.942); P < 0.001] and for a poor neurological outcome [AUC = 0.808; (95% CI 0.672–0.920) P = 0.038] in patients with minor intramedullary lesions (smaller than 36 mm). To further characterize the best critical age value associated with a poor functional and clinical prognosis, maximum Youden indexes were obtained for this subgroup of patients (those with lesions smaller than 36 mm). Maximum Youden indexes were obtained for an age of 57.3 and 54.7 in functional and clinical outcomes, respectively. Therefore, the critical value can be set at 55 years-old. In our sample, the sensitivity and specificity of this threshold to detect patients with a poor functional outcome was 100% and 67.5% (CI 57.8–79.7), respectively; while the sensitivity and specificity to detect those with a poor neurological outcome (complete SCI at 1-year-follow-up) was 100% and 74.5% (CI 64.8–84.5), respectively.
The results from this study demonstrate that elderly patients have a poorer functional and neurological outcome when the intramedullary lesion length is inferior to 36 mm. For larger lesions, intramedullary lesion length results to be the most dominant predicting factor, whereas the patient's age loses most of his predictive weight. In line with previous reports,, the present study also confirms that traumatic SCI in the older population was mostly secondary to minor falls, whereas traffic accidents are more frequent at a younger age. Preexisting degenerative cervical spine pathologies characteristics may well-explain why elderly patients frequently present with soft tissue injuries and long intramedullary lesions in the setting of low-energy traumatic events. Of note, we did not find significant differences in the neurological exam at presentation and at follow-up in older patients. This finding highlights the guarded neurorehabilitative potential in this older population.
To better delineate the true effect of age in the prognosis of patients with SCI after cervical trauma we preferred to apply objective measures such as functional and neurological scales. The WISCI was specifically developed with respect for gait assessment and walking recovery, which are poorly represented by other scales. Thus, some authors recommended to preferably use this latter scale as a principal indicator of functional status in SCI patients. For its simplicity, WISCI is also very reliable and consistent when data are pooled from medical records in retrospective studies. Likewise, AIS is an appropriate instrument to longitudinally evaluate SCI patients with respect to their motor and sensory function. The aforementioned assessment tools' validity and reliability were confirmed in many studies.,
The predicting value of patient age on functional capacity in chronic SCI, traumatic or not, has been previously studied.,The amount of papers that also tried to compare their results with radiological prognostic factors is, however, scarce. The recent development of the MRI and its new sequences has drawn a new scenario in which some of the radiological variables have a more prominent weight over other demographic variables.,, In this sense, it seems that the length of intramedullary lesion and the initial neurological function are the most powerful predicting factor in the acute phase following SCI.,, It is important to define the real significance of the patient's age in the outcome of patients with SCI. In the present study, after the cohort was stratified with respect to intramedullary lesion extension, we found out that age played a meaningful predicting role for lesion smaller than 36 mm, although the same factor lacked significant association with prognosis in the group of patients with larger lesions.
The relationship between the older age and poor outcome has already been described in other studies.,, What our work adds to the current body of literature is the statement that such a relationship is not as strong in patients with larger intramedullary lesions. In addition, we were able to determine a cutoff value for the patient's age associated with a worse prognosis (both neurological and functional), using ROC curves and the Maximum Youden Index. Various age-thresholds have been previously proposed, although determination of the cutoff value has not been commonly based on a proper statistical analysis, nor objective measurements. Our results are comparable to those reported by Franceschini et al. who performed a meritorious statistical analysis to establish a threshold of 50 years for a poor functional outcome. However, and despite using a method very similar to ours, the discriminatory capacity reported in their work is considerably lower than in the present study.
Strengths and limitations
In addition to the limitations inherent to the retrospective design of the study, the current work bears some potential shortcomings. The timing of MRI acquisition is a potential source of critical bias. Given dynamic changes occurring in the context of SCI, most authors have recommended performing an MRI within the acute phase to enhance its prognostic value.,, On the other hand, such strategy prevents the extrapolation of our results to minor cervical trauma, in which an MRI study is not warranted.
Age and the length of the intramedullary lesion are both independent prognostic factors during the acute phase of cervical SCI. An age cutoff of 55 years is critical to identify those patients at a higher risk of a poor functional and neurological outcome.
The authors would like to acknowledge the Writing Support Center from Western University who kindly proofread this article.
Declaration of patient consent
The authors certify that they have obtained all appropriate patient consent forms. In the form the patients have given their consent for their images and other clinical information to be reported in the journal. The patients understand that their names and initials will not be published and due efforts will be made to conceal their identity, but anonymity cannot be guaranteed.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
[Figure 1], [Figure 2]
[Table 1], [Table 2], [Table 3], [Table 4], [Table 5]