Analysis of motor and somatosensory evoked potentials during thoracic and thoracoabdominal aortic aneurysm repair
Article Outline
Objectives
Use of motor evoked potentials (MEP) and somatosensory evoked potentials (SSEP) monitoring during thoracic and thoracoabdominal aortic surgery is controversial. This study evaluated the intraoperative use of SSEP and MEP during thoracoabdominal repair and assessed their role in decreasing the risk of spinal cord ischemia and paralysis.
Methods
We conducted paired SSEP and MEP monitoring to assess agreement between the methods and their ability to predict neurologic outcome in 233 patients. Changes in SSEP and MEP monitoring were classified as no change, reversible change, or irreversible change during the intraoperative period and by the conclusion of surgery. Agreement between the methods was computed using the Cohen κ statistic. Sensitivity, specificity, and positive and negative predictive values were computed for each method on the immediate and delayed neurologic deficit.
Results
Immediate neurologic deficit, determined immediately upon awakening from anesthesia and confirmed by a neurologist, occurred in eight of 233 (3.4%) patients. For any change (reversible plus irreversible), agreement between MEP and SSEP was relatively low (κ = 0.53), despite being highly statistically significant (P < .001). MEP tended to overestimate SSEP for immediate neurologic deficit, demonstrating a 53% false-positive rate, compared with a 33% false-positive rate for SSEP (specificity ratio, 1.42; P < .0001). With irreversible change, agreement between the methods was 90% (κ = 0.896, P < .0001). Only irreversible change was significantly associated with neurologic outcome (odds ratio, 21.9; P < .00001 for SSEP; 60.8, P < .0001 for MEP), but sensitivity and positive predictive values were low (37% and 33% for SSEP; 22% and 45% for MEP, respectively). Reversible changes in neurophysiologic monitoring were not significantly associated with immediate neurologic deficit. Negative predictive values for all negative evoked potential findings were >98% for immediate deficit. No evoked potential variables were associated with delayed deficit.
Conclusions
SSEP and MEP monitoring were highly correlated only when intraoperative changes were irreversible. Reversible changes were not significantly associated with immediate neurologic deficit. Irreversible changes were significantly associated with immediate neurologic deficit, and the findings were identical for SSEP and MEP in this variable, indicating that the more complex MEP measures do not add further information to that obtained from SSEP. Normal SSEP and MEP findings had a strong negative predictive value, indicating that patients without signal loss are unlikely to awake with neurologic deficit.
Clinical Relevance
Monitoring of nerve activity during surgical procedures is vital to detect potential threats to neurologic integrity and to allow quick action to prevent deficits. This study is relevant because it provides outcome data comparing two ways to monitor evoked potentials during operations.
Thoracoabdominal aneurysm repair has evolved during the past 60 years. In the era of clamp and sew, paraplegia remained a major concern, with rates as high as 31%.1 However, this rate has currently declined precipitously with the use of adjuncts, including distal aortic perfusion, cerebrospinal fluid (CSF) drainage, sequential clamping, and reimplantation of intercostal arteries.2
Despite the improvement, paraplegia still remains a concern. This has led to approaches in predicting neurologic deficits (NDs) with the use of intraoperative neurophysiologic monitoring. In particular, cord function can be assessed with motor evoked potentials (MEP) and somatosensory evoked potentials (SSEP) as performed routinely in aortic surgery.3, 4 This study aims at evaluating and comparing neurophysiologic monitoring using SSEP and MEP during the repair of descending thoracic aortic (DTAA) and thoracoabdominal aortic aneurysm repair (TAAA).
Methods
Patients
The current study includes 233 patients (190 men and 43 women) with a mean age of 67 years (range, 18-87 years) who underwent DTAA (95) and TAAA (138) repairs between December 2004 and April 2008. We previously reported 176 of these patients in our series on the role of SSEP monitoring only.5 The 233 patients reported here underwent both SSEP and MEP monitoring intraoperatively.
Patient characteristics and the aneurysm extent breakdown are summarized in Table I, with extent I comprising 29 patients (12.5%), extent II, 25 (10.7%); extent III, 22 (9.4%); extent IV, 46 (19.7%); and extent V, 16 (6.9%). All data were collected prospectively as approved by the Committee for the Protection of Human Subjects at The University of Texas Houston Medical School. We defined ND as paraplegia or paraparesis upon awakening from the surgical procedure.
Table I. Patient characteristics
| Variable | No. (%) |
|---|---|
| All patients | 233 (100) |
| Age, y | |
| <65 | 59(25.3) |
| 65-70 | 60(25.7) |
| 71-76 | 57(24.5) |
| >76 | 57(24.5) |
| Sex | |
| Female | 43(19.0) |
| Male | 190(81.0) |
| COPD | |
| Yes | 103(44.2) |
| No | 130(55.8) |
| Coronary artery disease | |
| Yes | 63(27.0) |
| No | 170(73.0) |
| Aneurysm extent | |
| I | 29(12.5) |
| II | 25(10.7) |
| III | 22(9.4) |
| IV | 46(19.7) |
| V | 16(6.9) |
| Descending | 95(40.8) |
| GFR, mL/min/1.73 m2 | |
| ≤48 | 57(24.5) |
| 49-64 | 60(25.7) |
| 65-89 | 58(24.9) |
| ≥90 | 58(24.9) |
| Cross-clamp time, min | |
| <35 | 63(27.0) |
| 35-50 | 56(24.0) |
| 51-64 | 58(25.0) |
| >65 | 56(24.0) |
| Pump time, min | |
| ≤20 | 62(26.6) |
| 21-42 | 56(24.0) |
| 43-63 | 56(24.0) |
| ≥64 | 59(25.4) |
Anesthesia
In order not to interfere with SSEP and MEP monitoring, we have modified the anesthetic technique. Induction consists of front-end narcotic loading with fentanyl citrate (15 μg/kg), midazolam (0.05 mg/kg), propofol (0.5 mg/kg), and cis-atracurium (0.2 mg/kg). Patients are then maintained on a volatile agent (usually isoflurane) at 0.5 minimal alveolar concentration. Neuromuscular blockade is maintained at one-quarter train of four using supplemental doses of cis-atracurium. Hemodynamics are managed with a combination of volume adjustments, β-blockade, and dihydropyridines (nicardipine) as required. No compromise of neuromotor or neurosensory monitoring occurs with this approach.
At the end of the case, the patient is transitioned to a low-dose dexmedetomidine infusion (0.1-0.2 μg/kg/h) to maintain hemodynamic stability and provide sedative and analgesic control. It is important not to use the recommended loading dose of dexmedetomidine, because this is associated with profound bradycardia and hypotension. Minimal sedation is used because sedation makes identification of postoperative NDs difficult, and we know that early intervention is key to reversing delayed deficit.
Surgical technique
The details of our surgical technique have been described previously 2, 6, 7 and are briefly reviewed here. A thoracoabdominal incision is made extending towards the shoulder blade following the curve of the rib. The extent of the thoracoabdominal incision depends on extent of the aneurysm. The patient is then anticoagulated with heparin (1 mg/kg), followed by cannulation of the left lower pulmonary vein. This is attached to the centrifugal pump with an online heat exchanger, and the femoral artery is used to establish arterial inflow. The proximal descending thoracic aorta is isolated and lifted off the esophagus. We clamp sequentially starting at the middle descending thoracic aorta.
We suture an appropriately sized Dacron collagen or gelatin impregnated woven graft using 2-0 or 3-0 polypropylene suture in a running fashion. The distal clamp is then placed above the celiac axis, if possible, and the remainder of the thoracic aorta is opened. The lower intercostal arteries (8, 9, 10, 11, and 12) are identified, and if any are patent, a sidehole is created in the graft to reimplant the thoracic aorta containing the intercostal orifices to the graft. On rare occasions, we bypass to the individual intercostals arteries.
After this, the arterial pulsatile flow is restored to the implanted intercostal arteries. The graft is then passed into the aortic hiatus and into the abdominal cavity. If feasible, we clamp the infrarenal aorta or the iliac or the common femoral artery. The celiac, superior mesenteric, and the right and left renal arteries are inspected, and if atheromatous debris is present, endarterectomy is performed to that area. We perfuse and cool the celiac and superior mesenteric arteries with tepid blood and the renal arteries with special cold crystalloid solution to keep the kidney temperature <20°C. A sidehole is made opposite the visceral vessels and anastomosed to the graft using a 2-0 polypropylene suture. Once the anastomosis is completed, the patient is placed in head-down position, the aorta is flushed of all air and debris, and pulsatile flow is allowed into the visceral arteries.
Once the patient's nasopharyngeal temperature reaches >36°C, the patient is weaned from the distal aortic perfusion and all cannulae are removed. The patient is moved to the intensive care unit, where the neurologic status is monitored closely, keeping the CSF pressure <10 mm Hg for 3 days. When the patient is stable, extubated, and is able to move his or her extremities, the CSF catheter is removed on the third postoperative day. It is important to keep the mean arterial pressure under control and >80 mm Hg, and we like to keep the CSF drainage <10 mm Hg, and the hemoglobin level >10 mg/dL.
SSEP and MEP monitoring
A team led by a neurologist/neurophysiologist performed neurophysiologic monitoring. Eight-channel electroencephalogram monitoring was performed during the surgical procedure using a Viking IV (Nicolet, Madison, Wis). Evoked potentials were performed with the additional use of a Digitimer generator stimulator (Digitimer, Hertfordshire, United Kingdom) for the motor evoked potentials (Fig 1, Fig 2).
Fig 1.
Sensory evoked potential monitoring procedure.
Fig 2.
Motor evoked potential monitoring procedure.
For SSEP monitoring, stimulatory electrodes were placed bilaterally at the level of the malleolus. Recording electrodes were bilaterally placed at three levels: the popliteal fossa, cervical spine (C5), and vertex. The right and left posterior tibial nerves were stimulated at the ankle to obtain an averaged waveform that was repeated every 3 minutes continuously throughout the operation. The SSEP were bilaterally recorded at the three levels mentioned. A baseline SSEP tracing was obtained before the start of the operation. All subsequent tracing was superimposed and compared with this baseline. The traditional 10/50 rule was considered to define SSEP abnormalities. This is defined as a 10% change in latency or 50% change in amplitude. The evaluation of these three channels allowed us to distinguish SSEP changes related to spinal cord injury from peripheral nerve ischemia or cerebral injury.
For MEP monitoring, electrodes were placed at C3 and C4 as defined by the international 10-20 system stimulating the precentral gyrus. The stimulus consisted of train of five, an interstimulus interval of 2 milliseconds, and voltage as high as 400 V. Myogenic responses were recorded bilaterally with needle electrodes placed in the abductor digiti minimi, tibialis anterior, and abductor hallucis muscles. Compound muscle action potentials were checked intermittently throughout the operation and were considered on an all-or-none basis, present or absent.
Intraoperative corrective measures
During the operation, a series of corrective measures are instituted if SSEP or MEP findings show any signs of spinal cord dysfunction. These include increasing the mean central pressure to >80 mm Hg and increasing distal aortic pressure to >60 mm Hg. The anesthesia team will also lower the CSF pressure by free drainage and increase hemoglobin levels by transfusion. Furthermore, additional patent intercostal arteries are reimplanted, including T4 to T7 and L1. If the change occurs while the intercostal arteries are being implanted, the anastomosis is completed expeditiously to restore pulsatile flow to the intercostal arteries.
Statistical methods
Evoked potential measures were coded as permanent, transient, or any (permanent or transient) change. Permanent change was defined as absence of signal by the conclusion of surgery. Agreement between the methods (SSEP vs MEP) was estimated using the Cohen κ statistic. Screening test calculations (sensitivity, specificity, positive and negative predictive values, false-positive and false-negative rates) were performed using standard contingency table methods. Comparisons of screening test performance were computed using the Gart and Nam score with skewness correction. Associations between risk factors and outcome, and risk factors and SSEP changes were computed by contingency table methods using exact statistics. Computations were performed using SAS 9.1.3 software (SAS Institute, Cary, NC). Those not available in SAS were computed using locally produced programs, except the Gart and Nam score, which was computed using NCSS (Number Cruncher Statistical Systems, Kaysville, Utah).
Results
The ND rate in our series of patients was 3.4% (8 of 233). Neurologic deficit rate by extent is presented in Table II. Nine of 233 patients had permanent changes in SSEP, but only three (33%) exhibited NDs upon awakening. The other five of eight total patients with NDs did not have permanent SSEP changes. Similarly, for motor changes, 11 (4.7%) had permanent MEP defects, but only five (45%) went on to develop ND. The remaining three patients who sustained a ND did not have permanent MEP changes. Sensitivity for permanent SSEP and MEP changes was poor, at 37.5% and 62.5%, although specificity and negative predictive values were >97% for both evoked potential tests, with 2.2% and 1.4% deficit rates among patients with negative permanent changes in SSEP and MEP tracings, respectively. Agreement between SSEP and MEP permanent change findings was 89.6% by κ statistic (P < .0001).
Table II. Neurologic deficit by aneurysm extent
| Aneurysm extent | ||||||
|---|---|---|---|---|---|---|
| I | II | III | IV | V | Descending | |
| Patients, No. | 2 | 2 | 1 | 2 | 1 | 1 |
| Neurologic deficit, % | 6.9 | 8.0 | 4.55 | 4.35 | 0.0 | 1.05 |
Transient changes were not predictive of ND. For SSEP, three of the eight patients who were found to have NDs had transient changes. None of the MEP transient patients had NDs. Among the SSEP transient group, these three deficits represented 4% (3 of 68) of the exposed, which was not different from the overall 3.5% rate. That is, transient SSEP loss did not discriminate for ND risk.
For “any” change (reversible plus irreversible), sensitivity was improved but specificity worsened considerably. For SSEP, sensitivity increased to 75%, but specificity fell to 67%. Similarly for MEP “any” change, sensitivity was 88% but specificity was only 47%. The worsening of specificity with “any” change caused the specificities of the two methods to separate statistically (specificity ratio, 1.42; P < .0001), but this was not true for the improvement in sensitivity. Agreement for between MEP “any” change and SSEP “any” change was low (κ = 0.53) despite being statistically significant (P < .001). MEP tended to overestimate deficits relative to SSEP for this measure, demonstrating a 53% false-positive rate compared with a 33% false-positive rate for SSEP (P < .0001).
Only irreversible change was significantly associated with neurologic outcome, with odds ratios of 21.9 (P < .00001) for SSEP and 60.8 (P < .0001) for MEP. Reversible changes were not significantly associated with immediate ND. Negative predictive values for all negative evoked potential findings were >98% for immediate deficit. No evoked potential variables were associated with delayed deficit.
Discussion
In our current practice, the use of MEP and SSEP monitoring has become a standard part of intraoperative patient monitoring. This had led to our anesthesia team diligently modifying anesthetic agents to prevent interference with the neurophysiologic monitoring.
The role of SSEP has been evaluated in the past for DTAA and TAAA for its reliability in ruling out spinal dysfunction.5 Thus the question arises: What is the role of additional neurophysiologic monitoring with intraoperative MEP measurements? Many studies have shown the reliability of MEP monitoring in predicting postoperative ND.3, 4, 8
In this series of 233 patients, NDs developed in eight. Analysis of the SSEP monitoring revealed nine patients who had permanent change, of which NDs developed in only three. The remaining five patients with NDs did not have SSEP changes at the end of the operation. The sensitivity for this cohort of patients was low, at 37.5%; however, the specificity and negative predictive value remained high, indicating that patients with no permanent SSEP change at the end of the case had >97% probability of waking up without ND.
For MEP monitoring, 11 patients had permanent change but only five had immediate postoperative NDs. The remaining three patients with NDs did not have permanent change on MEP monitoring. Although the sensitivity increased with this monitoring, this was not statistically significant compared with SSEP monitoring. The specificity and negative predictive value remained comparable with SSEP, with a value >97%.
If we combined all changes intraoperatively (transient and permanent change), the sensitivity for both SSEP and MEP increase; however, this is at the cost of detecting more false-positives. Thus, the specificity of the test declined in order to detect more patients with a ND at the end of the case.
Neurophysiologic monitoring has allowed us to perform corrective measures when there is a change in signal. These include increasing central mean arterial pressure to >80 mm Hg, increasing distal aortic pressure to >60 mm Hg, decreasing CSF pressure by keeping the drain open, establishing pulsatile flow, and reattaching additional patent intercostals arteries (ie, T4 to T7 and L1). Despite our continuing effort to prevent NDs, the rate still remains at 3.4% in this series. Additional monitoring with MEP has not added any further benefit intraoperatively to help detect and perhaps change the likely course of NDs postoperatively. Whether we use SSEP or MEP, if there is a change during the operation, it is difficult to predict which patients will have NDs. However, we are able to predict with very high probability that if there is no change at the end of the case, the patient will wake up neurologically intact.
We retrospectively analyzed the prospectively collected MEP and SSEP data, thus setting up limitations to the study. Furthermore, this is not a randomized study in that we treated all neurophysiologic signal changes. Thus, the study did not include a group of patients where we decided not to treat with the corrective measures stated earlier when there was a signal change. This would be ethically impossible in our opinion as surgeons, when we know a portion of these patients upon awakening will have NDs and hence we have potentially altered ND in the remainder of these patients.
Although it might seem reasonable for us to perform extraordinary adjunctive maneuvers by increasing the distal perfusion pressures and implanting all intercostal arteries in all patients regardless of monitoring status, it may not be wise to do so. Increased distal pressure may come at the expense of adequate proximal perfusion pressure, and implanting all intercostal arteries may increase the operative time and bleeding risk. In almost all patients who demonstrate no changes in neurophysiologic monitoring, our standard protective measures are clearly adequate. In patients who do have changes, however, we believe that the additional margin of safety provided by further maneuvers justifies any increased risk associated with performing them.
It appears that sample sizes are required in these studies to be able to deduce whether permanent MEP change or SSEP change correlates with an immediate ND. The work of Jacobs et al3 has been used to show that MEP monitoring is a highly reliable tool to assess spinal cord function, such that patients who wake up with permanent loss of MEP had NDs and those with transient change had no immediate deficits.3 At this moment, the latter appears to be relatively comparable in our series, although the same information has been achieved by SSEP monitoring. This does not seem to apply in our series of 233 patients.
Conclusion
The role of neurophysiologic monitoring for DTAA and TAAA repair has continued to remain controversial. We have shown that SSEP monitoring is a reliable tool in ruling out ND. However, the addition of MEP monitoring has not contributed additional information in predicting ND.
Author contributions
References
- . Experience with 1509 patients undergoing thoracoabdominal aortic operations. J Vasc Surg. 1993;17:357–368
- Distal aortic perfusion and cerebrospinal fluid drainage for thoracoabdominal and descending thoracic aortic repair: ten years of organ protection. Ann Surg. 2003;238:372–380
- . The value of motor evoked potentials in reducing paraplegia during thoracoabdominal aneurysm repair. J Vasc Surg. 2006;43:239–246
- Usefulness of transcranial motor evoked potentials during thoracoabdominal aortic surgery. Ann Thorac Surg. 2007;83:456–461
- Role of somatosensory evoked potentials in predicting outcome during thoracoabdominal aortic repair. Ann Thorac Surg. 2007;84:782–787
- Determination of cerebral blood flow dynamics during retrograde cerebral perfusion using power M-mode transcranial Doppler. Ann Thorac Surg. 2003;76:704–709
- . Thoracic vasculature. In: Townsend CM, Beauchamp RD, Evers BM, Mattox KL editor. Sabiston textbook of surgery. 18th ed.. New York, NY: Elsevier; 2008;
- Spinal cord perfusion after extensive segmental artery sacrifice: can paraplegia be prevented?. Eur J Cardiothorac Surg. 2007;31:643–648
Competition of interest: none.
PII: S0741-5214(08)01333-5
doi:10.1016/j.jvs.2008.08.005
© 2009 The Society for Vascular Surgery. Published by Elsevier Inc. All rights reserved.
