The surgical treatment of spinal deformities or tumors continues to evolve with advanced surgical techniques of three-column procedures, such as complex spinal osteotomies, circumferential decompression of the spinal cord, or en bloc segmental resections of the vertebral column. These surgical procedures often present significant neurologic complications. The incidence of neurologic deficits in complex osteotomies was greater in simple spinal surgeries, such as single-level intervertebral fusions, at 0%–1% [1-5]. The reported incidence of neurologic complications during complex osteotomies for spinal deformity correction is 8%–27% [2,6,7]. In tumor resection surgeries, the incidence increases from up to 23.8%–65.4% [1,5,8]. These incidences represent a significantly higher risk than encountered during conventional spine surgeries, such as simple decompression with or without instrumented posterior fixation.

To improve neurologic safety, intraoperative neurophysiological monitoring (IONM) is used to detect and give a chance to reverse the related causes of neurologic complications during spinal surgery [9-13]. Many studies provide alarm criteria of the IONM signal change and confirm its reliability in detecting neurologic complications [14-17]. However, few of these studies reported a specific surgical procedure related to a significant IONM signal change during surgery. This study performed an in-depth analysis of the interplay between the surgical procedure and a significant IONM signal change, especially in three-column spinal surgery, to find a critical surgical step associated with a high risk of neurologic complications during surgery.

The Institutional Review Board of Seoul National University Hospital (IRB approval no., H-1703-156-840) and the Ethics Committee exempted the study for obtaining patient informed consent as this was a minimal risk study. All included surgeries were performed on 64 patients who underwent three-column spinal surgery from October 2011 to November 2015 by two spine-specialist orthopedic surgeons (H.K., B.S.C.) at one institution. Multimodality IONM data, including somatosensory-evoked potentials (SSEP) and motor-evoked potentials (MEP), were reviewed. Surgical procedures involving three columns included vertebral column resection (n=7), pedicle subtraction osteotomy (n=8), total en bloc segmental resection, such as spondylectomy and sagittal resection (n=12), piecemeal spondylectomy (n=10), and corpectomy via laminectomy with or without costotransversectomy (n=27). There were three cervical, 34 thoracic, and 31 lumbar levels. There were 27 females and 37 males. The mean age was 51.8 years. The diagnoses of the operated patients were congenital kyphoscoliosis (n=8), primary spinal tumor (n=8), metastatic spinal tumor (n=34), adult spinal deformity (n=7), trauma (n=5), and infection (n=2) (Table 1). The postoperative neurologic status of all included patients was followed for 36 at least months postoperatively. Patients received intravenous anesthetic agents (4.5 µg/mL propofol and 1.7 ng/mL remifentanil) compatible with IONM, and short-acting muscle relaxants were used for intubation but not during surgery. IONM was performed by an electroneurophysiology technologist under the supervision of a neurologist using the NIM-ECLIPSE Spinal System (Medtronic Sofamor Danek, Memphis, TN, USA) and following a set of monitoring protocols. MEP recording followed the application of transcranial electric stimulation. Transcranial anodal motor cortex stimulation was performed via electrical pulses through needle electrodes inserted over the C3’ and C4’ (1–2 cm anterior to C3 and C4 positions of the international 10–20 system). Trains of five biphasic pulses with an interstimulus interval of 1–2 ms were delivered at a rate of 1 Hz. Stimulation intensity ranged from 200 to 350 V. Monitored muscles were flexibly determined depending on the operation’s spinal level but generally included bilateral deltoid, biceps, thenar, tibialis anterior, gastrocnemius, and abductor halluces. Intraoperative real-time monitoring of electromyogram (EMG) was also performed at the same locations as those used in MEP monitoring. SSEPs were obtained by stimulating the median nerve at the wrist and the posterior tibial nerve at the ankle. A 200 μs square-wave electrical pulse was presented sequentially to the bilateral median and posterior tibial nerves at a rate of 2.31 Hz with stimulus intensity ranging from 20 to 40 mA. SSEP was recorded from needle electrodes placed on the scalp at locations C3 and C4 and Cz referenced to Fpz. The upper or lower extremities were selectively monitored depending on the operation’s spinal level. Baseline SSEP and MEP were obtained after anesthesia induction but before performing the surgical procedure. Peak-to-peak SSEP amplitude and latency were continuously monitored. Because MEPs provoke patient movement, trials were conducted only after confirmation or when requested during surgery. The request of MEP provocation was made by a surgeon in any recognition of potential insult to neurologic structures during surgery. Significant IONM signal changes were identified based on previously reported criteria. The criteria used to indicate an abnormal SSEP were either a >50% decrease in peakto-peak amplitude or a >10% increase in latency compared to the baseline. A reduction of more than 50% in MEP baseline amplitude elicited by direct cortical stimulation was considered a significant change. Relevant EMG activities reported to the surgeon included spikes, bursts, and trains [2,5,15-17]. Abnormal monitoring results were reported immediately by the electroneurophysiologist to allow the surgeon to suspend the surgery and investigate the cause in real-time. Technical and anesthetic variables were assessed and subsequently excluded as the cause of the significant change. Surgical variables, such as deformity correction and neural tissue adhesion or tethering, were examined. Corrective steps were performed when abnormal findings were found in these variables. Every event of a significant signal change was documented on the IONM record of each case in the context of surgical procedure and with comments on the suspicious cause through discussion with surgeons and neurologists.

The neurologic evaluation of each patient was conducted before surgery, within 24 hours postoperatively, and 36 months after surgery. The strength of the 10 key muscles ranged from 0 to 5. A follow-up was conducted up to 36 months after surgery in the group with an immediate postoperative motor deficit. Motor deficits that showed improvement in 36 months were considered transient, and motor deficits were considered permanent if they persisted after 36 months.

Data were imported into IBM SPSS ver. 20.1 (IBM Corp., Armonk, NY, USA) for analysis. Mann-Whitney test was used to analyze statistical differences between the two groups. Two-sided p-values <0.05 were considered statistically significant. The calculation of test operating characteristics (sensitivity and specificity) of the IONM was based on previous related studies [12,21,22]. Accordingly, IONM outcomes were classified in one of four ways. A true-positive event was defined as a signal degraded to warning criteria values, correlating with a postoperative neurologic deficit (whether or not the signal was recovered). A false-positive event was defined as a signal degraded to warning criteria values without postoperative neurologic deficit. The outcome was classified as true-negative when the signal remained consistent with baseline values, and the patient’s postoperative neurologic status was unchanged. Finally, a false-negative event was defined as a signal that remained consistent with baseline values throughout a surgical procedure, but the patient demonstrated postoperative neurologic deficit.

This study specifically reviewed 64 patients who underwent three-column spinal surgery with IONM and observed significant signal changes in 11 patients. Six of the 11 patients had postoperative neurologic deficits. The sensitivity of the IONM signal change was 100%, and the specificity was 91.4%. These results redemonstrated its usefulness and are comparable to other similar studies reporting the usefulness of IONM in various spinal surgeries [19,20,23-25]. In addition, this study demonstrated surgical causes related to the significant IONM signal change. The IONM events usually occurred when the spinal column was totally destabilized during surgery, increasing the risk of mechanical insult, such as contusion, translation, or tethering with or without epidural adhesion. However, it also occurred without any recognizable mechanical insult, while compromise of perfusion to the spinal cord was suspected, especially in excessive intraoperative bleeding or diminished vascular supply of the spinal cord. Many studies have reported that MEP is a sensitive marker for impending neurologic deficit due to spinal cord ischemia during spinal surgery and emphasized maintaining adequate blood pressure during surgery [12,21,26,27]. The average MAP measured at the time of signal change/loss of the patients presumed to have compromised perfusion was 45.7 mm Hg, statistically significantly lower than other patients (68.5 mm Hg, p<0.05). However, there was no statistically significant difference in EBL during surgery (p=1). Therefore, it can be argued that it is important to maintain proper blood pressure regardless of blood loss during surgery. It reinforces the opinion of previous related studies, and intraoperative MAP should be maintained to at least 60 to 65 mm Hg to minimize neurologic deficit [12,21,26].

In this study, 11 patients had significant signal changes that identified them as being at risk for neurologic injury during surgery. Immediate notification and close cooperation of the surgeon and the anesthesiologist, before permanent neurologic injury occurs, can result in adequate intervention, such as mechanical adjustment, administration of dexamethasone, and maintenance of adequate blood pressure. It is hard to prove the effectiveness of each intervention in each case; however, the IONM doubtlessly provides a motive to assess and avoid the potential cause of increasing neurologic complication risk by monitoring the neuroelectrophysiologic function of the spinal cord and detecting a significant change. In comparison to cases with the recovery of the IONM signal showing no permanent neurologic deficit, permanent neurologic deficits occurred in three patients, all with irreversible signal change until the end of the operation. Although it is not an absolute prognostic factor, recovery of the signal during surgery can be expected to have a favorable neurologic prognosis.

In this study, IONM signal change events were more common in tumor resection cases than deformity correction cases. Compared to 4.5% (1 of 22) of the patients without tumors, 23.8% (10 of 42) of the tumor patients had events and 14.3% (6 of 42) of the tumor patients had postoperative neurologic deficits (true-positive events). Several reasons were considered for the higher incidence rate of signal change or postoperative neurologic injury in tumor-related surgery. First, in surgical treatment for metastatic tumors, preoperative neurologic deficits are frequently present already, and there is often a high risk of neural tissue damage during tumor removal due to severe adhesion from preoperative radiation therapy. Second, in primary tumors where en bloc resection is usually performed with curative intention, significant structural instability is accompanied by the complete disconnection of the proximal and distal vertebral column to the segment of tumor location. Perfusion of the spinal cord is also likely diminished during circumferential dissection, 360° around the tumor to secure surgical margin.

There are several limitations to this study. First, this study has a relatively small number of subjects and is retrospective. However, given the rarity of three-column spinal surgery, these numbers are large enough to be meaningful even in retrospect. Second, sensory deficits were not checked, and this study focused only on postoperative motor function. Third, IONM data measurement may not have been accurate in patients with preoperative paralysis, as MEP response amplitudes in weak muscles are lower and more highly variable [28]. In addition, only patients who had significant change or loss of IONM signal were analyzed. This is an initial trial, and a comparison study to a control group would be required to increase credibility.

This study encompasses three-column spinal surgeries from the cervical spine to the lumbar spine. There were 11 cases (17.1%) of significant IONM signal change during surgery. The events usually occurred when the spinal column was totally destabilized during surgery in relation to mechanical insult such as contusion, translation, or tethering with or without epidural adhesion. However, it also occurred without recognizable mechanical insult, while compromise of perfusion to the spinal cord was suspected, especially in excessive intraoperative bleeding, or 360° dissection of the spinal cord with devascularization. Based on the results of this study, to enhance neurologic safety in three-column spinal surgery, surgeons should pay attention to protect the spinal cord from mechanical insult, especially when the spinal column was totally destabilized during surgery, and not to compromise perfusion to the spinal cord in close cooperation with a neurologist and anesthesiologist.