JBJS | Evaluation of Intraoperative Nerve-Monitoring During Insertion of an Iliosacral Implant in an Animal Model*

The Journal of Bone & Joint Surgery, Volume 81, Issue 11

Articles | November 01, 1999

Evaluation of Intraoperative Nerve-Monitoring During Insertion of an Iliosacral Implant in an Animal Model*

BERTON R. MOED, M.D.†; MICHAEL J. HARTMAN, D.V.M., M.D.‡; B. K. AHMAD, M.D.‡; DIANNA D. CODY, PH.D.‡; JOSEPH G. CRAIG, M.D.‡, DETROIT, MICHIGAN

View Disclosures and Other Information

Investigation performed at Henry Ford Hospital, Detroit

The Journal of Bone & Joint Surgery. 1999; 81:1529-37

5 Recommendations (Recommend) | 3 Comments | Saved by 3 Users Save Case

text A A A

Abstract

Background: The use of continuous electromyographic and somatosensory-evoked-potential monitoring systems has been advocated to assist in avoiding nerve-root injury during operations on the pelvic ring. More recently, it was suggested that stimulus-evoked electromyographic monitoring may further decrease the risk of iatrogenic nerve-root injury during posterior pelvic fixation by enabling the surgeon to determine the actual distance of an implant from a nerve root. The purpose of the current study was to evaluate the relative efficacy of these three methods of monitoring for minimizing the risk of neural injury during the placement of iliosacral implants.Methods: While the function of the first sacral nerve root was monitored with the use of stimulus-evoked electromyographic, continuous electromyographic, and somatosensory-evoked-potential monitoring techniques, a 2.0-millimeter stainless-steel Kirschner wire was progressively inserted, guided by a high-speed computerized tomographic scanner, into the first sacral body of seventeen hemipelves in nine dogs. The end point was contact with the nerve as demonstrated by the computerized tomographic images. It was expected that this end point would be heralded by a burst of spontaneous electromyographic activity and an abnormal somatosensory-evoked-potential signal. Anatomical dissection at the completion of the study documented the final position of the Kirschner wire.Results: Anatomical dissection demonstrated compression or penetration of the nerve root in sixteen of the seventeen specimens. A spontaneous burst of electromyographic activity was not recorded for any specimen on continuous electromyographic monitoring; this finding was significantly different from what had been expected (p < 0.001). Because of technical problems, somatosensory evoked potentials could be recorded for only twelve hemipelves that had nerve-root compression or penetration, and abnormal somatosensory evoked potentials were recorded for only one of the twelve; this finding was significantly different from what had been expected (p < 0.001). A total of 113 stimulus-evoked electromyographic data points were obtained. The correlation coefficient for the relationship between the current threshold recorded with stimulus-evoked electromyographic monitoring and the distance of the wire from the nerve was 0.801 (p < 0.001). The actual measured current thresholds were of an observed proportion not different from what had been expected (p = 0.48).Conclusions: Continuous electromyographic and somatosensory-evoked-potential monitoring techniques failed to indicate contact with the nerve root reliably in this animal model. However, stimulus-evoked electromyographic monitoring consistently provided reliable information indicating the proximity of the implant to the nerve root.Clinical Relevance: This investigation raises serious concerns regarding the validity of somatosensory-evoked-potential and spontaneous electromyographic monitoring for the purpose of minimizing nerve-root injury during the insertion of iliosacral implants. In addition, it further confirms the validity of the relationship between the stimulus-evoked electromyographic current threshold and the distance between the wire and the nerve and its potential applicability for nerve-monitoring.

Figures in this Article

Introduction

Introduction | Materials and Methods | Results | Discussion | References

Iliosacral screw fixation has emerged as one of the methods of choice for the operative stabilization of disruptions of the pelvic ring^7,19,20,29. Unfortunately, the ideal projected intraosseous course of the screw into the sacrum brings it in close proximity to the fifth lumbar nerve root, the sacral nerve roots, and the spinal canal^14,33. This operative technique may be particularly hazardous, even when performed by experts, as misdirection of the drill-bit, guide-wire, or implant may result in injury of a nerve root or the spinal canal. The prevalence of iatrogenic neural injury has been reported to be as high as 14 percent (three of twenty-two patients who had insertion of thirty iliosacral screws)³⁵.

Nerve-monitoring techniques, such as the monitoring of somatosensory evoked potentials and continuous electromyography, have been used in an attempt to minimize the risk of iatrogenic neural injury^11,35. However, these monitoring systems record abnormal signals only after a neural injury has altered a previously normal signal (somatosensory-evoked-potential monitoring) or has generated an abnormal spontaneous burst of neurotonic activity (continuous electromyography). Therefore, with either method, the signal is associated with some magnitude of neural injury caused by traction, crushing, laceration, or thermal damage. This injury may be minor and reversible (such as that caused by a guide-pin merely pressing on or tenting the nerve) or it may be major and permanent (such as a laceration or crush injury)²². A proactive monitoring system is much more desirable. Reports of both false-positive and false-negative findings with use of these monitoring techniques are cause for additional concern regarding their intraoperative reliability^10,23,27.

Intraoperative stimulus-evoked electromyographic monitoring consists of the determination of a minimum current threshold that is required to evoke an electromyographic response; theoretically, this current threshold correlates directly with the distance of an implant from the nerve structure at risk. This method has been used successfully during operations involving the facial and recurrent laryngeal nerves^2,9,18,25,28 and during pedicle-screw fixation of the lumbosacral spine in both animal studies and clinical series^{1,4,5,16,17,32}. More recently, stimulus-evoked electromyographic monitoring was shown to be technically feasible in operations on the pelvic ring^23,24. It is thought that this method of nerve-monitoring may decrease the risk of iatrogenic nerve-root injury by providing an early warning system that advises the surgeon of impending neural injury, thereby allowing the direction of the wire, drill-bit, or screw to be altered in order to avoid contact with neural structures^23,24. The purpose of the current study was to evaluate the efficacy of stimulus-evoked electromyographic monitoring compared with that of continuous electromyographic and somatosensory-evoked-potential monitoring for minimizing the risk of neural injury during the placement of iliosacral implants.

*No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article. Funds were received in total or partial support of the research or clinical study presented in this article. The funding sources were the Department of Orthopaedic Surgery Research and Chairman's Funds.

†Department of Orthopaedic Surgery, University Health Center, Suite 7C, 4201 St. Antoine Boulevard, Detroit, Michigan 48201.

‡Departments of Orthopaedics (M. J. H.), Neurology (B. K. A.), and Radiology (D. D. C. and J. G. C.), Henry Ford Hospital, 2799 West Grand Boulevard, Detroit, Michigan 48202.

†Department of Orthopaedic Surgery, University Health Center, Suite 7C, 4201 St. Antoine Boulevard, Detroit, Michigan 48201.

‡Departments of Orthopaedics (M. J. H.), Neurology (B. K. A.), and Radiology (D. D. C. and J. G. C.), Henry Ford Hospital, 2799 West Grand Boulevard, Detroit, Michigan 48202.

View Large | Download Slide(.ppt)
Add to My JBJS

Figs. 1-A through 1-D: Representative post-processed computerized tomographic scans of Dog 9, made with use of a GE High Speed Advantage scanner (General Electric Medical Systems, Milwaukee, Wisconsin) that produced one-millimeter contiguous axial sections. Fig. 1-A: Axial computerized tomographic section for the fourth data point on the left side. The method of measurement, which revealed the indicated distance of 3.3 millimeters, is shown.

View Large | Download Slide(.ppt)
Add to My JBJS

Fig. 1-B Reformatted oblique computerized tomographic reconstruction for the same data point as in Fig. 1-A. In this instance, the measured distance (3.3 millimeters) was identical to that obtained with the axial measurement.

View Large | Download Slide(.ppt)
Add to My JBJS

Fig. 1-C Axial computerized tomographic section (Fig. 1-C) and reformatted oblique reconstruction (Fig. 1-D) for the fourth data point on the right side. In this instance, the distance measured from the axial section (6.9 millimeters) differed by 0.2 millimeter from that obtained with the oblique reconstruction (7.1 millimeters). The numerical average (7.0 millimeters) was used for correlation with the nerve-monitoring observations.

View Large | Download Slide(.ppt)
Add to My JBJS

Fig. 1-D Axial computerized tomographic section (Fig. 1-C) and reformatted oblique reconstruction (Fig. 1-D) for the fourth data point on the right side. In this instance, the distance measured from the axial section (6.9 millimeters) differed by 0.2 millimeter from that obtained with the oblique reconstruction (7.1 millimeters). The numerical average (7.0 millimeters) was used for correlation with the nerve-monitoring observations.

View Large | Download Slide(.ppt)
Add to My JBJS

Fig. 2 Representative specimen, from Dog 5, demonstrating the anatomical dissection and direct visualization of the Kirschner wires and the nerve roots. One Kirschner wire compressed and stretched the first sacral nerve root (right), and another impaled it (left). The remaining distal nerve roots in the cauda equina are seen in the midline.

View Large | Download Slide(.ppt)
Add to My JBJS

Fig. 3 Graph showing the association between the stimulus-evoked electromyographic current threshold (in milliamperes) and the computerized tomographic measurements of the distance (in millimeters) between the Kirschner wire and the nerve. Regression analysis performed with use of a generalized estimating equation demonstrated a correlation coefficient of 0.801, indicating a very strong linear relationship between the measurement of distance and the current threshold (p < 0.001). Dashed lines = 95 percent confidence interval, and solid line = estimated regression line.

View Larger

TABLE I SUMMARY OF RESULTS

*Signals adequate for monitoring could not be obtained.

Dog	Side	Final Anatomical Position of Kirschner Wire	Final Current Threshold (mA)	Abnormal Spontaneous Electromyographic Response	Abnormal Somatosensory Evoked Potentials
1	L	Compressing nerve	2.4	No	No
2	R	Compressing nerve	1.3	No	No
	L	Compressing nerve	1.0	No	No
3	R	Compressing nerve	1.6	No	No
	L	Compressing nerve	1.4	No	No
4	R	Compressing nerve	5.9	No	*
	L	Compressing nerve	2.0	No	*
5	R	Compressing nerve	1.5	No	No
	L	Penetrating nerve	2.4	No	No
6	R	Penetrating nerve	2.7	No	No
	L	Compressing nerve	1.0	No	No (34% decrease)
7	R	Compressing nerve	4.3	No	*
	L	Penetrating nerve	3.3	No	*
8	R	1 mm away, just outside of S1 canal	6.3	No	No
	L	Penetrating nerve	1.7	No	Yes (50% decrease)
9	R	Compressing nerve	1.7	No	No
	L	Compressing nerve	1.3	No	No

Add to My JBJS

TABLE I SUMMARY OF RESULTS

*Signals adequate for monitoring could not be obtained.

Dog	Side	Final Anatomical Position of Kirschner Wire	Final Current Threshold (mA)	Abnormal Spontaneous Electromyographic Response	Abnormal Somatosensory Evoked Potentials
1	L	Compressing nerve	2.4	No	No
2	R	Compressing nerve	1.3	No	No
	L	Compressing nerve	1.0	No	No
3	R	Compressing nerve	1.6	No	No
	L	Compressing nerve	1.4	No	No
4	R	Compressing nerve	5.9	No	*
	L	Compressing nerve	2.0	No	*
5	R	Compressing nerve	1.5	No	No
	L	Penetrating nerve	2.4	No	No
6	R	Penetrating nerve	2.7	No	No
	L	Compressing nerve	1.0	No	No (34% decrease)
7	R	Compressing nerve	4.3	No	*
	L	Penetrating nerve	3.3	No	*
8	R	1 mm away, just outside of S1 canal	6.3	No	No
	L	Penetrating nerve	1.7	No	Yes (50% decrease)
9	R	Compressing nerve	1.7	No	No
	L	Compressing nerve	1.3	No	No

Materials and Methods

Introduction | Materials and Methods | Results | Discussion | References

A canine model was selected for study^13,15,24, and approval was obtained from the Care of Experimental Animals Committee. A power study was performed prospectively to determine the number of hemipelves required to provide 90 percent power, assuming that measurements would be obtained from both sides of the same animal, with a minimum of four data measurements (with stimulus-evoked electromyography) obtained for each side⁶. As the result of this analysis, experiments were conducted on seventeen hemipelves obtained from nine large (average weight, sixty pounds [27.2 kilograms]) adult mongrel dogs.

Each dog was placed prone on a GE High Speed Advantage computerized axial tomographic scanner (General Electric Medical Systems, Milwaukee, Wisconsin) fifteen minutes after being sedated. General anesthesia was induced with intravenous administration of propofol (4.5 milligrams per kilogram of body weight), followed by endotracheal intubation. Anesthesia was maintained with use of 1.0 percent isoflurane inhalation supplemented by an intravenous bolus of fentanyl (0.002 to 0.006 milligram per kilogram), administered every thirty minutes. This anesthetic combination was selected in order not to attenuate the somatosensory-evoked-potential and electromyographic signals during the operative procedure^13,15,24. The animals were monitored continuously by one of us (M. J. H.) who is a veterinarian. In this way, factors such as temperature, blood pressure, and level of anesthesia were kept constant.

While the function of the first sacral nerve root was monitored with use of the Viking IV System (Nicolet Biomedical Instruments, Madison, Wisconsin) and previously described stimulus-evoked electromyographic, continuous electromyographic, and somatosensory-evoked-potential monitoring techniques^{12,13,15,23,24,35}, a 2.0-millimeter stainless-steel Kirschner wire was inserted progressively from the external surface of the ilium, across the sacroiliac joint, and into the first sacral body of the hemipelvis toward the first sacral canal and the first sacral nerve root with use of the computerized tomographic scanner for control of direction. The end point was defined as contact with the nerve root as demonstrated by the computerized tomographic images. It was expected that this end point would be heralded by an abnormal spontaneous neurotonic burst of electromyographic activity^23,35, an abnormal somatosensory-evoked-potential signal (defined as a decrease in amplitude of more than 50 percent or an increase in latency of at least 10 percent^22,27), and a stimulus-evoked electromyographic current threshold of four milliamperes or less^5,16,17,24. It also was expected that broaching of the neural canal without actual nerve contact would coincide with a stimulus-evoked electromyographic current threshold of six milliamperes or less, no abnormal spontaneous electromyographic activity, and no alteration of the normal somatosensory-evoked-potential monitoring pattern^{5,16,17,23,24}.

The constant-current method of stimulus-evoked electromyographic monitoring^5,17,23,24 was used, with the introduced 2.0-millimeter Kirschner wire acting as an electrode. Monopolar, monophasic square-wave stimulation was performed at three hertz with a duration of 0.2 millisecond and the cathode (stimulating) lead attached to the Kirschner wire. An electroencephalographic-type needle electrode placed in the subcutaneous tissue in the midline served as the anode (reference electrode). Electroencephalographic-type recording needle electrodes (monitoring electrodes) were placed in the gastrocnemius muscle to allow determination of stimulus-evoked electromyographic thresholds as well as continuous monitoring of spontaneous electromyographic activity for the first sacral nerve root^3,21,23,24. A searching current of forty milliamperes was initially applied to the Kirschner wire. The stimulus-evoked electromyographic current threshold was defined as the lowest current needed to evoke an electromyographic response of twenty-microvolt amplitude in the gastrocnemius muscle²⁴. Two somatosensory-evoked-potential recording needle electrodes were placed midline in the subcutaneous tissue overlying the skull, one cranial and one caudal. A somatosensory-evoked-potential stimulating needle electrode was placed near the tibial nerve in both hindlegs¹³.

The GE High Speed Advantage computerized tomographic scanner was capable of a one-millimeter slice thickness with a spatial resolution of 0.45 millimeter (according to clinical compliance inspection data obtained from Annual Compliance Reports of the Division of Physics, Department of Diagnostic Radiology at Henry Ford Hospital). The computerized tomographic scans were made at 140 kilovolts peak, 300 milliamperes, and one second, with a fifteen to twenty-centimeter display field of view, a standard algorithm, and an axial scan mode.

In order to maximize the accuracy of the measurements of distance, two methods were used. First, the distance between the tip of the Kirschner wire and the center of the nerve-root foramen was estimated to the nearest millimeter during scanning in the axial views and was recorded. Generally, a series of ten images centered around the tip of the Kirschner wire were made for each measurement step. The entire set of images (approximately 200 images for each dog) was transferred to a workstation for more detailed analysis after image acquisition was complete. A GE Advantage Windows workstation, running version-1.2 software, was used to determine the distance between the tip of the Kirschner wire and the center of the nerve-root foramen. The center of the foramen was used as a landmark, as noted earlier, as it was always visible even when streak artifact from the Kirschner wire was present. The image that was used for the initial estimation of approximate distance during data acquisition was reanalyzed with use of more precise calipers and was magnified to a thirteen-centimeter field of view. The distance from the tip of the Kirschner wire to the center of the foramen was determined (Fig. 1-A). Then, each set of ten images, which represented a measurement set, was loaded in the reformat program on the workstation. All images were magnified to a thirteen-centimeter field of view, and the vertebrae window settings were applied to allow better visualization of the Kirschner wire. Second, an oblique cut plane that followed the center of the pin was generated. This oblique image showed a complete section of the Kirschner wire in which the tip was clearly visible, as was the foramen. The distance between the tip of the Kirschner wire and the center of the foramen was measured, on the oblique image, with use of the position of the cursor shown on the axial image to help to position the end of the measurement vector in the center of the foramen. This distance, which in most cases was the same as or within a few tenths of a millimeter of that measured on the axial image, was recorded for each measurement step (Fig. 1-B). The distance function of the workstation provides measurements to the nearest tenth of a millimeter. For each measurement step, the numerical average of the distances (to the nearest tenth of a millimeter) measured on the oblique and axial images was used for correlation with the nerve-monitoring observations (Figs. 1-C and 1-D).

Computerized tomographic targeting of the first sacral nerve root and foramen, with progressive advancement of the Kirschner wire toward this target, was performed in a specific sequence. First, the first sacral foramen was visualized on a scout view and the first sacral nerve root was identified within its confines, with the center of the foramen used as a landmark. Potential starting points on the skin then were marked radiographically. Second, another scan was made to identify the appropriate insertion point that would place the Kirschner wire in line with the foramen, with the Kirschner wire, the nerve root, and the foramen visualized in the same computerized tomographic plane. With use of a small mallet (to eliminate any possibility of heat generation and thermal injury as a confounding variable), the 2.0-millimeter stainless-steel Kirschner wire then was inserted into the ilium, across the sacroiliac joint, and into the cortex of the sacrum.

Once the Kirschner wire was in the sacral cortex and was visualized on the axial computerized tomographic image in the same plane as the first sacral nerve root, progressive insertion of the Kirschner wire was begun. More than one trial of this sequence of insertion was needed in order to place the wire at the sacral cortex in line with and in the same plane as the targeted first sacral nerve root. The first recording of distance was made at this point, along with recordings of the results of simultaneous stimulus-evoked electromyographic, continuous electromyographic, and somatosensory-evoked-potential monitoring of the function of the first sacral nerve root. The Kirschner wire then was slowly advanced several millimeters. The new distance from the nerve root (estimated with the computerized tomographic scanner), the new stimulus-evoked electromyographic minimum current threshold, and any abnormal somatosensory-evoked-potential or continuous electromyographic activity then were recorded in a data table. Four to nine measurements of distance (depending on the size of the animal) and the corresponding observations of nerve function were recorded for each of the seventeen hemipelves. Each operative procedure concluded with advancement of the tip of the Kirschner wire into the first sacral foramen and presumed contact with the nerve root as demonstrated by the computerized tomographic images. At the end of the operative procedure, each dog was killed and the actual location of the Kirschner wire was documented with anatomical dissection and direct visualization of the Kirschner wire and the first sacral nerve root (Fig. 2). In the absence of direct contact, the distance from the first sacral nerve root to the nearest part of the Kirschner wire was measured with calipers. Statistical analysis of these measurements was provided by the Department of Biostatistics and Research Epidemiology at the Henry Ford Health Sciences Center.

Results

Introduction | Materials and Methods | Results | Discussion | References

A total of 113 data points were obtained to determine the relationship between the stimulus-evoked electromyographic current threshold and the distance between the Kirschner wire and the nerve. In eight dogs, both hemipelves were used, generating two sets of data points from the same animal. Since independence cannot be assumed among the measurements obtained for each dog, the generalized estimating equation was used in a regression analysis⁸. Essentially, use of the generalized estimating equation involves calculation of a weighted average of the simple linear regression lines obtained from the nine dogs but also accounts for the fact that two separate hemipelves were sampled in eight of the nine animals. The resulting generalized estimating equation regression equation was: distance = -0.093 + (0.359) (current). The corresponding correlation coefficient was 0.801, with a root mean square error of 2.47, indicating a highly significant (p < 0.001) regression fit (Fig. 3). There was a narrow 95 percent confidence band for the line of linear regression (dashed lines, Fig. 3), indicating high precision.

Anatomical dissection performed at the completion of the study revealed that sixteen of the seventeen Kirschner wires had fully penetrated the first sacral canal and were in contact with the nerve root. Four wires had actually penetrated the nerve root, and twelve were compressing the root but had not penetrated its substance. The remaining Kirschner wire, which was embedded in the cortical bone of the foramen, was just short of penetrating the canal and was measured as being one millimeter from contact with the nerve root. This was the only wire associated with a final stimulus-evoked electromyographic current threshold of more than six milliamperes (6.3 milliamperes); this finding is consistent with the hypothesis that six milliamperes or less is indicative of penetration of the canal at the first sacral level. Fourteen of the remaining sixteen current thresholds were less than four milliamperes (Table I); this observed proportion was not significantly different from the expected finding of sixteen of sixteen (Fisher exact test, p = 0.48).

A spontaneous neurotonic burst of electromyographic activity was not recorded for any specimen. This finding was significantly different from the rate of sixteen of sixteen (a spontaneous burst for sixteen of the sixteen Kirschner wires) that was expected in the nerve-contact group (Fisher exact test, p < 0.001). Because of technical difficulties of undetermined origin, it was not possible to obtain somatosensory-evoked-potential signals that were satisfactory for monitoring during the procedures on two dogs (four Kirschner wires). Of the remaining thirteen Kirschner wires, twelve were in contact with the nerve root; three wires had penetrated the nerve root, and nine were compressing it. The somatosensory-evoked-potential signal was abnormal for only one of these twelve wires (Dog 8L, Table I). (This signal, which had a 50 percent decrease in amplitude, was categorized as abnormal even though it did not meet the criterion of more than a 50 percent decrease.) This finding was significantly different from the expected finding that twelve of the twelve wires that were in contact with the nerve root would be associated with an abnormal signal (Fisher exact test, p < 0.001). This marginally abnormal somatosensory evoked potential occurred in association with a Kirschner wire that had penetrated the nerve root. Although categorized as normal, a second somatosensory evoked potential (Dog 6L, Table I), occurring in association with a Kirschner wire that was compressing the nerve root, approached the abnormal range (a 34 percent decrease in amplitude). The remaining ten Kirschner wires were associated with final somatosensory-evoked-potential signals that were essentially unchanged from baseline. Although two nerves were analyzed in eight of the nine dogs, the assumption of independence was accepted for these Fisher exact-test comparisons because of the descriptive nature of this analysis.

Discussion

Introduction | Materials and Methods | Results | Discussion | References

The insertion of iliosacral screws is challenging because of the narrow path that the implant must follow in order to remain intraosseous and the proximity of important structures in the surrounding soft tissues³². As little as 4 degrees of change in the angulation of the surgeon's hand as he or she inserts the screw can cause a major and potentially disastrous alteration in the trajectory of the implant³³. Since the desired course of the screw cannot be visualized directly during the operative procedure, c-arm fluoroscopy with use of multiple views has become the standard method for determining the position of the implant^{14,19,20,29,30}. However, accurate placement of the screw with use of this method relies on obtaining an anatomical reduction and on radiographic imaging of osseous landmarks, which occasionally may be obscured or misinterpreted^26,30,35.

The utilization of continuous electromyographic and somatosensory-evoked-potential monitoring systems has been advocated to help the surgeon to avoid nerve-root injury during the fixation of pelvic fractures^11,35. These monitoring methods provide information by indicating that there has been some change from a normal to an abnormal condition. The usefulness of somatosensory-evoked-potential monitoring is limited by the need for a rapid response to an abnormal signal and by the fact that the sensory rather than the motor pathway is being monitored²². Continuous electromyographic monitoring relies on the observation of one or more spontaneous bursts of neurotonic activity, indicating that there has been some mechanical insult to the nerve root^12,35. Therefore, both of these monitoring techniques provide information regarding an acute adverse neurological event after the fact. The hope is that the loss of signal (somatosensory-evoked-potential monitoring) or the occurrence of an abnormal signal (continuous electromyography) has been caused by a reversible situation, such as stretching of the nerve by retractors or minor nerve-root compression^11,12,22. Laceration or crushing of the neural structures by a guide-pin, drill-bit, or implant cannot be prevented with use of either of these after-the-fact monitoring techniques. Also, reports of both false-positive and false-negative results are cause for additional concern regarding the intraoperative reliability of these techniques^10,23,27.

Intraoperative stimulus-evoked electromyographic monitoring is of interest because of its potential to provide information indicating the proximity of the neural structure being monitored. Ideally, this method would indicate that an instrument or implant is getting too close to the nerve at risk. This advance warning should allow a change in course to avoid any contact with or injury of the nerve. The method has proved to be effective in operations involving the facial and recurrent laryngeal nerves, in which the objective is to locate but not injure a neural structure that is obscured by abnormal anatomy^2,9,18,25,28. Early indications are that a modification of this technique is helpful for avoiding neural injury during placement of pedicle screws in the lumbosacral spine^{1,4,5,16,17,32} and during insertion of iliosacral screws for fixation of pelvic fractures^23,24.

Following the lead of previous investigations^{11,23,24,34,35}, we sought to further evaluate the efficacy of stimulus-evoked electromyographic, continuous electromyographic, and somatosensory-evoked-potential monitoring of neural function during the placement of iliosacral implants. A canine model was selected for study because it was used similarly in a previous study²⁴, the anatomy of the canine pelvis is comparable with that of the human pelvis²¹, and a large, available animal was needed. The first sacral nerve root was chosen as the target nerve root because it is easy to locate radiographically in the first sacral foramen and because of its applicability to the clinical setting. In the dog, the major nerve-root contributions to the sciatic nerve arise from the sixth and seventh lumbar and first and second sacral levels^3,21. The first sacral nerve root is the main component of the tibial nerve and thereby provides the major innervation to the gastrocnemius muscle^3,21. Therefore, in this animal model, only the gastrocnemius muscle needs to be monitored (with electromyography) and only the tibial nerve needs to be stimulated (with somatosensory-evoked-potential monitoring) in order to approximate the clinical setting for detection of injury of the first sacral nerve root. Although one of us (B. R. M.) and colleagues²⁴ used an initial stimulus-evoked electromyographic searching current of twenty milliamperes in an earlier study, we chose a current of forty milliamperes in the present study. This higher current was selected because we sought to elicit initial current thresholds at a greater distance from the nerve root, from as far lateral as the sacral side of the sacroiliac joint.

Our previous studies have indicated that computerized tomographic measurement of distance is associated with problems both in the clinical setting²³ and in the dog model²⁴. In our previous clinical study²³, the axial scans consisted of three-millimeter contiguous sections; the cut planes of these axial scans were in the approximate plane of insertion of the screw but were oblique to the sacral foramina. In addition, beam-hardening artifact from the large amount of implanted stainless steel was appreciable. In our previous study of dogs²³, beam-hardening artifact was especially relevant because of the relatively numerous (four) and large (2.5-millimeter) Kirschner wires that had been implanted in the hemipelvis. In the present study, adjustments were made to minimize these problems. The use of a single 2.0-millimeter Kirschner wire in each hemipelvis reduced implant-induced artifact to a minimum. Application of multiple one-millimeter contiguous sections at a high-voltage setting (140 kilovolts peak) to identify the tip of the Kirschner wire, meticulous initial localization of the Kirschner wire in line with the first sacral foramen, and incorporation of oblique reconstructions in the exact plane of the Kirschner wire and the first sacral foramen were performed in an attempt to reduce measurement error. However, one aspect of this computerized tomographic-based study design, by its very nature, created a certain margin of error. Our method was to estimate the location of the nerve root with the assumption of a constant relationship to the center of the neural foramen. The center of the first sacral foramen, which was assumed to correspond to the center of the nerve root, was used for all measurements. In our previous, anatomical dissection-based study²⁴, the recorded measurement consisted of the distance from the tip of the Kirschner wire to the closest aspect of the nerve root. Considering the relatively small overall size of the first sacral nerve root and the small discrepancy between the size of the first sacral foramen and the first sacral nerve root, the margin for error related to this part of the study design was thought to be acceptably small. As testimony to the accuracy of this technique, sixteen of the seventeen attempts at targeting the first sacral nerve root proved successful and the one errant Kirschner wire was only one millimeter from contact with the nerve root. Additional evidence of the success of the approach might be the experimental finding of a correlation of 0.801 between the stimulus-evoked electromyographic current threshold and the distance between the wire and the nerve, which is not dissimilar from the value of 0.94 in our previous animal study; that study, as noted earlier, involved direct measurements of distance after anatomical dissection of dog pelves²⁴. However, our indirect (rather than direct) method of targeting the center (rather than the nearest aspect) of the nerve root may help to explain any difference from the findings of the previous study as well as the two aberrant final current thresholds of 4.3 and 5.9 milliamperes in the present study that were higher than the level (four milliamperes or less) expected to indicate nerve contact.

In the current study, we found a strong correlation between the stimulus-evoked electromyographic current threshold and the distance between the wire and the nerve; this finding was consistent with that of many other investigations^4,5,16,17,24 and, therefore, it was not unexpected. What is surprising is the essentially complete inability of continuous electromyographic and somatosensory-evoked-potential monitoring to identify an injured first sacral nerve root in this animal model. In fact, the single abnormal somatosensory-evoked-potential recording actually was marginally normal, but classifying it as abnormal was considered to be the most conservative and appropriate approach. We did not anticipate these findings, despite previous reports of false-negative results^10,27, as it is unclear how a nerve root could be injured without generating a spontaneous neurotonic burst of electromyographic activity or an adverse effect on the somatosensory evoked potentials. One explanation is that the nerve roots were not actually injured in this model despite the contact with the Kirschner wire. We did not perform histological examination of the nerve-root specimens to try to determine whether there was permanent damage.

We did not consider obtaining histological specimens because of the unreliability of the association between histological findings and nerve function. In the absence of gross anatomical disruption, nerve conductivity can be restored before histological changes associated with nerve block are corrected, there may be nerve block without any detectable modification of nerve morphology, and nerve fibers with a grossly modified sheath structure may continue to function efficiently³¹. Therefore, histological examination of these specimens actually would have added confusion to the study. A possible flaw in our study design is that we did not allow the animals to recover from the anesthetic and then attempt to document injury by performing a physical examination. However, it is very difficult to imagine a nerve root retaining normal function after it has been either impaled or compressed and stretched by a relatively large implant to the extent that was seen in this study (Fig. 2). Certainly, similar positioning of an implant in the clinical situation is highly undesirable. A more likely explanation is that these so-called reactive monitoring techniques are not sensitive enough to recognize damage to a relatively small component of the neural structure that is being monitored. The first sacral nerve root makes up only a portion of the tibial nerve (the structure that was being monitored), and only a portion of the nerve root (potentially a very small portion, when the nerve root was impaled by the sharp implant) was damaged. Therefore, it is possible that a more generalized insult to a larger structure (such as stretch or compression of the sciatic nerve during operative treatment of an acetabular fracture or a similar injury to the lumbosacral plexus during reduction of a displaced pelvic fracture) is amenable to continuous electromyographic and somatosensory-evoked-potential monitoring, whereas injury of a nerve root caused by insertion of an iliosacral implant is not.

It should be emphasized that the objective of nerve-monitoring during insertion of an iliosacral implant is to avoid any injury, including neurapraxia. Only by avoiding nerve contact can this objective be ensured. Nerves are sensitive to both pressure and stretch. The magnitude of the initial insult is extremely important, but so is the duration³¹. Stretching of or pressure on the nerve root, as applied in the current study (and in the clinical situation), may result in neurapraxia or may result, not in any discernible acute injury, but in an injury over time^22,31. In the clinical setting, insertion of the drill-bit or Kirschner wire is followed by insertion of a much larger implant (a screw two to three times its size). If the smaller implant is in contact with the nerve (even if it has not caused great injury to the nerve), the large threaded implant that is inserted subsequently certainly will cause problems. Ideally, monitoring should be sensitive enough to detect acute nerve contact, regardless of how slight or reversible the initial injury is. Some level of injury must have occurred to the nerve roots in the current study, four of which were impaled and twelve of which were severely stretched and compressed. Perhaps the injury was slight or reversible, or both; however, it certainly was not detectable with two of the nerve-monitoring techniques. If these two methods had been used in the clinical setting, the insult would have continued and would have been exacerbated by the larger implant that would have been inserted subsequently.

This study raises serious concerns regarding the validity of somatosensory-evoked-potential and spontaneous electromyographic monitoring for the purpose of minimizing nerve-root injury during the insertion of iliosacral implants. In addition, it further confirms the validity of the relationship between stimulus-evoked electromyographic current thresholds and the distance between implants and nerves and the potential applicability of these thresholds for nerve-monitoring in the clinical setting. However, it is important to recognize that stimulus-evoked electromyographic monitoring assumes a baseline of normal neurological function. Preoperative neural abnormalities, such as a preexisting neuropathy or (more commonly) a neural injury caused by the trauma of a fracture, alter the response of the nerve to the stimulus of a given current³².

NOTE: The authors thank Gary Chase, Ph.D., Suzanne Havstad, M.S., and Gordon Jacobsen, M.S., for assistance in providing the statistical analysis.

References

Introduction | Materials and Methods | Results | Discussion | References

Ahmad, B. K.; Szkiladz, E.; Roth, H.; Easton, R.; and Jacobson, G.: Individual nerve root current threshold measurement during pedicle screw placement. Read at the Annual Meeting of the American Society of Neurophysiological Monitoring, Chicago, Illinois, May 5, 1994.

Benecke, J. E., Jr.; Calder, H. B.; and Chadwick, G.: Facial nerve monitoring during acoustic neuroma removal. Laryngoscope,97: 697-700, 1987.97697 1987 [PubMed]

Bennett, D.: An anatomical and histological study of the sciatic nerve, relating to peripheral nerve injuries in the dog and cat. J. Small Animal Pract.,17: 379-386, 1976.17379 1976

Calancie, B.; Lebwohl, N.; Madsen, P.; and Klose, K. J.: Intraoperative evoked EMG monitoring in an animal model. A new technique for evaluating pedicle screw placement. Spine,17: 1229-1235, 1992.171229 1992 [PubMed]

Calancie, B.; Madsen, P.; and Lebwohl, N.: Stimulus-evoked EMG monitoring during transpedicular lumbosacral spine instrumentation. Initial clinical results. Spine,19: 2780-2786, 1994.192780 1994 [PubMed]

Cohen, J.: Statistical Power Analysis for the Behavioral Sciences. Rev. ed. New York, Academic Press, 1977.

Cole, J. D.; Blum, D. A.; and Ansel, L. J.: Outcome after fixation of unstable posterior pelvic ring injuries. Clin. Orthop.,329: 160-179, 1996.329160 1996 [PubMed]

Diggle, P. J.; Liang, K.-Y.; and Zeger, S. L.: Analysis of Longitudinal Data, pp. 143-145. New York, Oxford University Press, 1995.

Gantz, B. J.: Intraoperative facial nerve monitoring. Am. J. Otol., Supplement, pp. 58-61, 1985.

Helfet, D. L.; Hissa, E. A.; Sergay, S.; and Mast, J. W.: Somatosensory evoked potential monitoring in the surgical management of acute acetabular fractures. J. Orthop. Trauma,5: 161-166, 1991.5161 1991 [PubMed]

Helfet, D. L.; Koval, K. J.; Hissa, E. A.; Patterson, S.; DiPasquale, T.; and Sanders, R.: Intraoperative somatosensory evoked potential monitoring during acute pelvic fracture surgery. J. Orthop. Trauma,9: 28-34, 1995.928 1995 [PubMed]

Helfet, D. L.; Anand, N.; Malkani, A. L.; Heise, C.; Quinn, T. J.; Green, D. S.; and Burga, S.: Intraoperative monitoring of motor pathways during operative fixation of acute acetabular fractures. J. Orthop. Trauma,11: 2-6, 1997.112 1997 [PubMed]

Holliday, T. A.; Weldon, N. E.; and Ealand, B. G.: Percutaneous recording of evoked spinal cord potentials of dogs. Am. J. Vet. Res.,40: 326-333, 1979.40326 1979 [PubMed]

Kellam, J. P.; McLaren, A.; and Tile, M.: Internal fixation. In Fractures of the Pelvis and Acetabulum, edited by M. Tile. Ed. 2, pp. 150-199. Baltimore, Williams and Wilkins, 1995.

Kornegay, J. N.; Marshall, A. E.; Purinton, P. T.; and Oliver, J. E., Jr.: Somatosensory-evoked potential in clinically normal dogs. Am. J. Vet. Res.,42: 70-73, 1981.4270 1981 [PubMed]

Lenke, L. G.; Padberg, A. M.; Russo, M. H.; Bridwell, K. H.; and Gelb, D. E.: Triggered electromyographic threshold for accuracy of pedicle screw placement. An animal model and clinical correlation. Spine,20: 1585-1591, 1995.201585 1995 [PubMed]

Maguire, J.; Wallace, S.; Madigan, R.; Leppanen, R.; and Draper, V.: Evaluation of intrapedicular screw position using intraoperative evoked electromyography. Spine,20: 1068-1074, 1995.201068 1995 [PubMed]

Maloney, R. W.; Murcek, B. W.; Steehler, K. W.; Sibly, D.; and Maloney, R. E.: A new method for intraoperative recurrent laryngeal nerve monitoring. Ear, Nose, and Throat J.,73: 30-33, 1994.7330 1994

Matta, J. M., and Saucedo, T.: Internal fixation of pelvic ring fractures. Clin. Orthop.,242: 83-97, 1989.24283 1989 [PubMed]

Matta, J. M., and Tornetta, P., III: Internal fixation of unstable pelvic ring injuries. Clin. Orthop.,329: 129-140, 1996.329129 1996 [PubMed]

Miller, M. E.: Miller's Anatomy of the Dog, edited by H. E. Evans and G. C. Christensen. Ed. 2, pp. 107-225, 972-1029. Philadelphia, W. B. Saunders, 1979.

Moed, B. R.; Maxey, J. W.; and Minster, G. J.: Intraoperative somatosensory evoked potential monitoring of the sciatic nerve: an animal model. J. Orthop. Trauma,6: 59-65, 1992.659 1992 [PubMed]

Moed, B. R.; Ahmad, B. K.; Craig, J. G.; Jacobson, G. P.; and Anders, M. J.: Intraoperative monitoring with stimulus-evoked electromyography during placement of iliosacral screws. An initial clinical study. J. Bone and Joint Surg.,80-A: 537-546, April 1998.80-A537 1998

Moed, B. R.; Anders, M. J.; Ahmad, B. K.; Craig, J. G.; and Jacobson, G. P.: Intraoperative stimulus-evoked electromyographic monitoring for placement of iliosacral implants: an animal model. J. Orthop. Trauma,12: 85-89, 1998.1285 1998 [PubMed]

Moller, A. R., and Jannetta, P. J.: Preservation of facial function during removal of acoustic neuromas. Use of monopolar constant-voltage stimulation and EMG. J. Neurosurg.,61: 757-760, 1984.61757 1984 [PubMed]

Mostafavi, H. R., and Tornetta, P., III: Radiologic evaluation of the pelvis. Clin. Orthop.,329: 6-14, 1996.3296 1996 [PubMed]

Nash, C. L., Jr., and Brown, R. H.: Current concepts review. Spinal cord monitoring. J. Bone and Joint Surg.,71-A: 627-630, April 1989.71-A627 1989

Rice, D. H., and Cone-Wesson, B.: Intraoperative recurrent laryngeal nerve monitoring. Otolaryngol.—Head and Neck Surg.,105: 372-375, 1991.105372 1991

Routt, M. L. C., Jr., and Simonian, P. T.: Closed reduction and percutaneous skeletal fixation of sacral fractures. Clin. Orthop.,329: 121-128, 1996.329121 1996 [PubMed]

Routt, M. L., Jr.; Simonian, P. T.; Agnew, S. G.; and Mann, F. A.: Radiographic recognition of the sacral alar slope for optimal placement of iliosacral screws: a cadaveric and clinical study. J. Orthop. Trauma,10: 171-177, 1996.10171 1996 [PubMed]

Sunderland, S.: Nerves and Nerve Injuries. Ed. 2, pp. 73-74. New York, Churchill Livingstone, 1978.

Szkiladz, E.; Calder, H. B.; and Easton, R. W.: Intraoperative lumbo-sacral nerve root threshold measurement. Read at the Annual Meeting of the American Society of Neurophysiological Monitoring, San Francisco, California, May 18, 1995.

Templeman, D.; Schmidt, A.; Freese, J.; and Weisman, I.: Proximity of iliosacral screws to neurovascular structures after internal fixation. Clin. Orthop.,329: 194-198, 1996.329194 1996 [PubMed]

Vrahas, M.; Gordon, R. G.; Mears, D. C.; Krieger, D.; and Sclabassi, R. J.: Intraoperative somatosensory evoked potential monitoring of pelvic and acetabular fractures. J. Orthop. Trauma,6: 50-58, 1992.650 1992 [PubMed]

Webb, L. X.; De Araujo, W.; Donofrio, P.; Santos, C.; Zaccaro, D. J.; and Haygood, T.: Continuous EMG monitoring for placement of percutaneous iliosacral screws. Orthop. Trans.,20: 134, 1996.20134 1996

Topics

View Large | Download Slide(.ppt)
Add to My JBJS

View Larger

TABLE I SUMMARY OF RESULTS

*Signals adequate for monitoring could not be obtained.

Dog	Side	Final Anatomical Position of Kirschner Wire	Final Current Threshold (mA)	Abnormal Spontaneous Electromyographic Response	Abnormal Somatosensory Evoked Potentials
1	L	Compressing nerve	2.4	No	No
2	R	Compressing nerve	1.3	No	No
	L	Compressing nerve	1.0	No	No
3	R	Compressing nerve	1.6	No	No
	L	Compressing nerve	1.4	No	No
4	R	Compressing nerve	5.9	No	*
	L	Compressing nerve	2.0	No	*
5	R	Compressing nerve	1.5	No	No
	L	Penetrating nerve	2.4	No	No
6	R	Penetrating nerve	2.7	No	No
	L	Compressing nerve	1.0	No	No (34% decrease)
7	R	Compressing nerve	4.3	No	*
	L	Penetrating nerve	3.3	No	*
8	R	1 mm away, just outside of S1 canal	6.3	No	No
	L	Penetrating nerve	1.7	No	Yes (50% decrease)
9	R	Compressing nerve	1.7	No	No
	L	Compressing nerve	1.3	No	No

Add to My JBJS

TABLE I SUMMARY OF RESULTS

*Signals adequate for monitoring could not be obtained.

Dog	Side	Final Anatomical Position of Kirschner Wire	Final Current Threshold (mA)	Abnormal Spontaneous Electromyographic Response	Abnormal Somatosensory Evoked Potentials
1	L	Compressing nerve	2.4	No	No
2	R	Compressing nerve	1.3	No	No
	L	Compressing nerve	1.0	No	No
3	R	Compressing nerve	1.6	No	No
	L	Compressing nerve	1.4	No	No
4	R	Compressing nerve	5.9	No	*
	L	Compressing nerve	2.0	No	*
5	R	Compressing nerve	1.5	No	No
	L	Penetrating nerve	2.4	No	No
6	R	Penetrating nerve	2.7	No	No
	L	Compressing nerve	1.0	No	No (34% decrease)
7	R	Compressing nerve	4.3	No	*
	L	Penetrating nerve	3.3	No	*
8	R	1 mm away, just outside of S1 canal	6.3	No	No
	L	Penetrating nerve	1.7	No	Yes (50% decrease)
9	R	Compressing nerve	1.7	No	No
	L	Compressing nerve	1.3	No	No