IONM has been widely used in spine surgery, its sensitivity and specificity vary in different surgery type. However, it remains unknown how the multi-modal IONM performs in TTK corrective surgery. The present study aimed to identify the performance and utility of multi-modal IONM of SSEP and MEP in TTK corrective surgery.
Our study showed that multi-modal IONM with SSEP and MEP can effectively indicate a potential neural injury and predict a PND during TTK corrective surgery. A sensitivity of 0.75 and a specificity of 0.86 for multi-modal IONM alerting predicted a PND, which indicated a good performance. Based on the alerting, surgeons are sensitive to notice an impending spinal cord impingement and consequently reverse it. Interestingly, 66.7% cases could be rescued from an alerting, whereas one third would develop PNDs. One having an IONM alerting has a chance of 25% to develop a PND. While, if no alerting occurs, it is 98% to be safe. Incidence of PND in alerting cases was 17-fold higher than those without alerting. Furthermore, AUC of 0.884 showed an excellent accuracy of EP recovery in diagnosing PND, which can lead to a precise evaluation of prognosis.
Many studies had shown single EP monitoring (MEP or SSEP) was less sensitive to predict a neurological deficit. When single SSEP and MEP were separately analyzed, the sensitivity of them were 0.33 and 0.67 (both < 0.75) respectively, which is in consistence with previous studies[16, 18, 20, 21]. 242 lumbar PSO cases with single MEP were analyzed by Darryl Lau et al., calculating a sensitivity of 30%[16]. Similarly, in a study of single SSEP monitoring during thoracic fusion and decompression surgery, 39 of 771 had significant potential change under a criterion of >50% reduction in amplitude or prolongation of > 10% in latency, generating a low sensitivity of 19% and a specificity of 96%[18]. A multi-modal IONM is regarded to improve diagnostic accuracy. There is a popularity in performing multi-modal IONM including SSEP, MEP, D-wave and EMG in spine surgery[22]. In compliance with Martin Sutter’s research, when the alerting criterion was defined as any relevant change indicating potential neurological injury in single monitoring parameters, multi-modal IONM performs better than single IONM, with a sensitivity of 93% versus 13-81%. The sensitivity and specificity were a little higher than that in ours, 93% vs 75% and 99.1% vs 85.9% respectively[23].However, in his study, 2728 patients with different spine pathologies were included compared to our only TTK patients. Mixed pathologies and their mild alerting criteria generated the difference. In a meta-analysis comprising 2,052 patients with idiopathic scoliosis, its reported sensitivity and specificity were 76.5% and 95.1%, which were comparable to our study. Surgery for idiopathic scoliosis is similar to TTK involving high-risk procedures of screw-implant and correction[24]. With aggressive alarm criteria, an excellent sensitivity of 100% and specificity of 89.3% were achieved when multi-modal IONM with SSEP and MEP were performed in thoracic decompressive surgery[25]. Though, the efficacy of multi-modal IONM varies in different surgery type, it performs better than single modal IONM. However, it is not the more modalities are used, the better IONM performs. In a study of IONM with MEP, SSEP and EMG in adult spinal deformity. The multi-modal sensitivity in patients with osteotomy was 67%, specificity 98%. Similarly, a study of IONM for intradural extramedullary spinal tumors resection with MEP, SSEP and D-wave had sensitive of 85.7% and specificity of 97%[26].
The following factors were identified to influence on IONM alerting: greater preoperative kyphosis angle, more osteotomy levels, longer operation time, more bleeding, three-column osteotomy, static tuberculosis and titanium mesh use. The logistic regression identified that severe preoperative kyphosis (≥80°) was an independent risk factor for IONM alerting, rather than other factors above. The OR of active tuberculosis between alerting group and no alerting group was 0.16, which indicated lower risk of neural injury in active tuberculosis. Active tuberculosis cases usually have a minor kyphosis angle and less rigid deformity compared to static ones. Titanium mesh was usually used in static tuberculosis to take place of resected vertebrae. Greater preoperative kyphosis requires more osteotomy levels, longer operation time and more bleeding. Three-column osteotomy increases risk for neural injury. Kamerlink[27] studied 60 cases with spinal sagittal plane deformity monitored by SSEP and MEP, in which 13 PSOs (21.67%) were performed. They observed 5 (8.33%) IONM alerting under their criteria of a 10% increase in latency and/or 50% decrease in amplitude of SSEPs, or absence of MEPs. Despite their stricter alerting criteria, only 13 PSOs were involved. Though, A higher incidence of alerting was observed in our study including 10 PVCRs (14.71%) and 2 PSOs (2.94%), the three-column osteotomy rate in present study was 54.41% (37/68). No alerting occurred in SPO/Ponte osteotomy cases.
The only one false-negative (25%) was characterized by a low amplitude at baseline. The patient had progressive decreased muscle strength of lower extremities, which was deteriorated in four days from grade 4 (Medical Research Council grading system) of both lower extremities in admission to grade 3 (left) and grade 2 (right) before surgery. After surgery, the muscle strength decreased to grade 2 of both sides and did not reverse. Unfortunately, the strength of right lower extremity turned down to grade 0 before discharge. In a previous report of risk factors for failure of MEP baseline, muscle strength lower than grade 3 are associated with baseline failure as well as poor outcome[28]. Our false-negative case had a similar situation with the study. Some inherent injuries possibly existed before surgery. Strict assessment of preoperative neural function should be achieved to evaluate value of surgery and outcome of neural function. This study [29] concluded that infeasible IONM occurs when preoperative MEP and SSEP are not recordable, which also supports the theory above. In a retrospective spine surgery cohort analysis of 62,038 patients, 109 patients were identified with PND, in which 22 (20.2%) cases were absent of IONM alerting (false negative). Of the 22 false negative cases, 3 (13.6%) were failed to detect an interpretable IONM baseline and an IONM change[30]. Though, we established an interpretable baseline in our false negative case, its amplitude was very low. Therefore, in those with preoperative decreased neural function, if no interpretable baseline is obtained, it equals an IONM alerting.
There were several potential drawbacks of the present study. First, the retrospective nature and relatively small sample size, as well as incomplete or inaccurate documentation, might lead to bias. Specific grade of muscle strength and specific sensory disorder were not analyzed to minimize bias. Second, a PND was defined as neural function deterioration from pre- to postoperation where the neural examination was manual. Subjectivity existed in both examinee and examiner. What’s more, EMG and D-wave modalities were absent in our center. Future investigations should be aimed at determining whether multi-modal monitoring, additional EMG and D-wave, improves specificity for detection of PND. Prospective randomized controlled trials with a multi-center design and larger number of TTK patients are required to further establish performance characteristics and diagnostic accuracy of IONM in TTK.