Median sternotomy with symmetric sternal retraction has been thought to produce brachial plexus injury, especially after high sternal retractor placement with excessive opening [1-3]. The hands-up (HU) position, in which the arms are behind the head and elevated above the level of the table, is thought to reduce brachial plexus injury during symmetric sternal retraction [4,5]. Studies attempting to correlate brachial plexus injury to arm position during open heart surgery have yielded conflicting results [6,7], possibly because the methods to detect nerve injury lacked sensitivity and/or the determinations were done after the traumatic insult had occurred. In addition, subclinical injury to the nerve, which may be as high as 87%, may have gone undetected [8,9].
Asymmetric sternal retraction for internal mammary artery (IMA) harvest may also increase the risk of damage to the brachial plexus [10-11]. Using somatosensory evoked potentials (SSEPs), an increased degree of deterioration in nerve conduction of the brachial plexus occurs when asymmetric retraction is applied to the sternum as compared to symmetric retraction [12,13].
This study was performed to test the role of arm position (either adducted at the side [AAS] or abducted hands-up [HU]) on reducing stress to the brachial plexus during asymmetric sternal retraction. SSEPs were monitored as a surrogate for clinical examination of the brachial plexus during anesthesia.
After approval by our institutional review board, 80 patients undergoing elective coronary artery bypass grafting (CABG) with dissection and use of a left IMA bypass graft were admitted to the study. On the day prior to the surgery a detailed neurologic assessment, specifically related to the brachial plexus, was performed by a nurse practitioner blinded to the intraoperative portion of the study. This same examination was done within 2 days of tracheal extubation, which was performed within 24 h of surgical completion. All motor and sensory functions of the brachial plexus were examined along with a thorough history regarding upper extremity pain, parathesias, numbness, and weakness. Patients were excluded from the study if neurologic dysfunction or deficits were noted during the preoperative examination. Patients were also excluded if they had a history of polyneuritis secondary to diabetes, transient ischemic attacks, or syncopal episodes or were more than 150% ideal body weight.
On the day of surgery, patients were randomly assigned to one of two groups of 40 patients each. Patients in Group 1 underwent CABG with their arms adducted and placed at their side. Elbows were protected with foam pads and hands were placed with thumbs in the upright position. Group 2 patients had their CABG performed with their arms in the HU position (Figure 1). This position is maintained by placing foam wedges under the shoulders to prevent posterior displacement of the humerus and to keep the elbows 20 cm above the level of the operating room table. Arms are abducted no more than 90 degrees and the elbows are flexed and padded with foam. Tape is placed to hold the arm without compression of the ulnar nerve at the elbow. IMA harvest was performed, always on the left side, using either a Pittman[TM] or Rultract[TM] retractor which was alternated for each consecutive patient. The Pittman[TM] retractor (Minnesota Scientific Instruments Inc., Minneapolis, MN) consists of two vertical bars suspended from a horizontal bar. One end of each vertical bar is placed in the chest cavity and attached to the sternum. The other end is attached to the horizontal bar. The left side of the sternum is opened asymmetrically by a screw mechanism at the top of the vertical bar that attaches to the horizontal stabilizing bar. The Rultract[TM] sternal retractor (Rultract Inc., Cleveland, OH) has two small vertical arms attached to a short broad horizontal beam. This beam connects to a cable that runs to a winch assembly attached at the top of a large vertical pole that is fixed to the table. The chest is opened asymmetrically by inserting the two small vertical arms underneath the left side of the sternum and using the winch assembly to open the chest.
Before arrival in the operating room all patients were medicated with morphine 0.1 mg/kg intramuscularly, scopolamine, 0.4 mg intramuscularly, and lorazepam, 1 mg, orally. All patients were monitored with an electrocardiogram, automated blood pressure cuff, mass spectrometer, pulse oximeter, esophageal stethoscope, temperature probe, intraarterial catheter, and pulmonary artery or central venous catheter always inserted through a right internal jugular approach. The blood pressure cuff was always placed on the right arm approximately 5 cm above the ulnar groove while the arterial catheter was always placed in the left radial artery. After preoxygenation, anesthesia was induced with fentanyl, 40 to 70 micro g/kg intravenously. Pancuronium 0.1 mg/kg was used to facilitate intubation. Oxygen 100% was used throughout the procedure with isoflurane 0.2 of 0.4% end-tidal, if needed. Hemodynamic variables were kept within 15% of baseline with the use of intravenous phenylephrine for hypotension and nitroprusside for hypertension. Body temperature throughout the pre bypass study period was maintained within 0.5 degrees C of preinduction temperature.
Cortical SSEP were monitored using the Nicolet Pathfinder[TM] MEGA eight-channel electrodiagnostic system (Nicolet Biomedical Instruments, Madison, WI) by a trained nurse blinded to both arm position and retractor. Before induction, fine-gauge platinum electrodes (Grass) were placed subdermally and used for both stimulating and recording responses in both arms. The median and ulnar nerves were stimulated at the wrist with peripheral responses recorded at Erb's point. Subcortical responses were monitored at C2, and cortical responses monitored at C3 prime, CY prime with an Fpz ground as defined by the international 10-20 system . Standard stimulating and recording variables consisted of a stimulus current of 15-20 mA with an evoked response induced by 250 to 300 stimulus repetitions. Filters were set at 30-1000 Hz and impedances were less than 2 k Omega. The monitor uses asynchronous parallel stimulation at a rate of 7.1 Hz. Latency and amplitude measurements for the monitoring session were determined from the N19 peak and the P22 trough. A baseline (preincision) set of SSEP recordings were obtained in all patients after induction of anesthesia with their arms at their sides. These recordings served as the baseline for comparisons in Group 1. After the arms were positioned HU, another preincisional set of SSEP recordings were obtained which were then used as the baseline measurement for all subsequent readings in Group 2 patients. SSEPs were continuously monitored throughout the case until the initiation of cardiopulmonary bypass with particular attention paid to SSEP waveform changes at the following time points: incision, asymmetric retractor placement in the chest cavity, retractor removal, and after symmetric sternal retractor (Cooley) placement. Factors that affect SSEP waveforms were minimized; isoflurane was used at concentrations less than 0.5% end-tidal and core body temperature was maintained within 0.5 degrees C of preinduction values until cooling was initiated for bypass.
Demographic data were compared between groups using Student's t-test. Total anesthesia (time patient entered operating room to intensive care unit admission), bypass (from initiation to separation from bypass), and aortic cross-clamp (time of clamp application to removal) times along with the number of internal jugular cannulation attempts were also compared in like manner. Intragroup SSEP changes in latency and amplitude during asymmetric retractor insertion, removal, and placement of symmetric sternal retractor were compared to preincision values using repeated measures multiple analysis of variance. Intergroup SSEP comparisons, at the specific time points monitored but with different arm positions, were also analyzed using repeated measures multiple analysis of variance. Intergroup comparisons of the number of subjects who had SSEP amplitude changes of >50% and who had postoperative neurologic symptoms were performed using chi squared analysis. A P value of <0.05 was considered significant for all statistical models used. All values are represented as mean +/- SEM.
A small difference in age and height was noted between the groups (Table 1). Intraoperative total anesthesia, CABG and aortic cross-clamp times as well as the number of internal jugular cannulations attempts were not different between groups (Table 1).
Changing upper extremity position from the AAS to the HU position in Group 2 produced minimal, nonsignificant changes in SSEP waveform latency on the right and left sides (21.9 +/- 0.2 ms vs 21.9 +/- 0.2 ms and 21.6 +/- 0.3 ms vs 21.7 +/- 0.3 ms, respectively) while amplitudes decreased significantly on both the right and left sides (2.21 +/- 0.29 micro V vs 2.02 +/- 0.26 micro V and 1.67 +/- 0.20 micro V vs 1.51 +/- 0.17 micro V [P < 0.05], respectively).
In patients who underwent asymmetric sternal retraction for IMA harvest with either the AAS or HU positioning, SSEP latency was unchanged from initial values with retractor placement. No differences were noted in SSEP latency changes between the HU versus the AAS groups when either the Pittman or Rultract retractors were compared separately.
SSEP waveform amplitudes on the left side decreased from baseline values in both Groups 1 and 2 when the asymmetric retractor was used to expose the left IMA (Table 2). A similar decrease, but of less magnitude, was noted on the right side. Decreases in amplitude returned toward baseline after retractor removal (Table 2). No significant intergroup mean differences were noted between the HU and the AAS positions (Table 2).
Most patients in both groups experienced decreases in SSEP amplitude on the left side (90%-95%) after asymmetric retractor placement; 85%-95% of the patients also had amplitude decreases on the right side. Although more patients in Group 1 (AAS) compared to Group 2 (HU) experienced decreases of greater than 50% in SSEP waveform amplitude on the left side after asymmetric retractor placement, these differences did not reach significance (Table 3). Incidence of right-sided amplitude decreases of 50% or more was less than that observed on the left side in both groups (Table 3). Comparison of the two different retractors, without regard to arm positioning, showed that patients who used the Pittman device had a 33% (13 of 40) incidence of SSEP amplitude decreases of >50% with retractor placement compared to a 53% (21 of 40) incidence in the Rultract group (P = 0.07).
Of the seven patients in Group 1 with complaints of brachial plexus injury postoperatively, six had leftsided symptoms (Table 3). These included three patients with an ulnar nerve distribution of symptoms (fourth and fifth digit numbness and parathesias at elbow and median forearm) and three with upper plexus or median nerve sensory complaints (shoulder pain, numbness, or tingling noted in middle and index finger and pain in antecubital area of arm). The one patient with right-sided symptoms had median and radial nerve involvement. In Group 2, four patients had complaints with three of these patients having left-sided symptoms (Table 3). In all these patients median or upper plexus involvement was noted without ulnar nerve symptoms. In all clinically symptomatic patients, an SSEP amplitude decrease of greater than 50% was noted on the side where plexopathy was found. In these patients, waveform amplitudes recovered more slowly and did not fully return to incisional levels prior to discontinuation of monitoring. One additional patient in Group 1 complained of right arm pain during placement of the right internal jugular catheter, but had no postoperative neurologic symptoms. Contrasting the roles of the two sternal retractors with both arm positions revealed no real benefit for either device. With the Rultract retractor, 12 of 20 patients had > 50% decreases in SSEP amplitudes from baseline during retractor placement with AAS positioning, while 9 of 20 had similar decreases with HU positioning. The subgroup in which the Pittman retractor was used with AAS positioning had a similar incidence of substantial SSEP amplitude decreases (8 of 20); with the HU positioning, 5 of 20 experienced this amplitude decrease.
The results of this study confirm that asymmetric retractor placement for IMA harvest does affect the integrity of the brachial plexus. As previously demonstrated, changes primarily in amplitude of brachial plexus SSEP waveforms do occur after IMA retractor placement which are suggestive of nerve injury [12-13]. This study found the incidence of clinical plexopathy after IMA harvest to be approximately 14%, a value within the reported range of 12%-37.5% [1,2,6,7]. The present study is unique in examining the effect of arm position on neurophysiologic determinants of brachial plexus integrity during IMA harvest. It demonstrated that the HU position did not have an appreciable effect in reducing nerve stretch or compression injury as demonstrated by changes in SSEP waveforms. Our data do show, however, that patients who were positioned HU had a lower incidence of substantial SSEP amplitude decreases (>50%), a degree of change often associated with neural damage . In addition, when the Pittman retractor was used in association with HU positioning, the lowest incidence of substantial SSEP changes was observed, which correlated with the absence of brachial plexus symptoms in this subgroup of patients.
Vander Salm et al.  postulated that nerve injury was not produced by stretching of the brachial plexus during sternal retraction, but resulted from penetration of the nerves by a fractured first rib that occurred after sternotomy. The incidence of this occurrence was noted to be almost 50% . We believe a less traumatic stretch or compression injury to the nerve must occur, since nerve penetration would produce a permanent impairment and possible laceration, which would not be as reversible as noted by the return of SSEP responses after retractor removal.
Not all stretch or compression injuries will result in postoperative clinical manifestations. Marganitt et al.  noted that 87% of patients had postoperative subclinical electromyogram findings of brachial plexopathy after symmetric sternal retraction with the Ankeney retractor, but none complained of symptoms. Our data support this finding; 90-95% of our patients experienced neurophysiologic evidence of potential brachial plexus injury during asymmetric sternal retraction, although a much smaller percentage experienced symptoms postoperatively. Hickey et al.  also demonstrated that there are variable degrees of neurologic deficit after asymmetric sternal retraction with the Favalaro retractor. They postulated that normal or improving SSEP waveforms, as noted in many of our patients after retractor removal, indicate minimal peripheral nerve dysfunction that will improve within the intraoperative period. In fact, right-sided SSEP amplitudes returned more rapidly after retractor removal, which suggests that the degree of injury to the nerve that occurred on the left (retractor) side was greater than that which occurred on the side opposite the retractor. The appearance of some right-sided SSEP changes and clinical plexopathy is not surprising; such changes are seen with symmetric retraction and both asymmetric retractors used here produce some right-sided distortion, as well.
Tomlinson et al.  believed that, during symmetric sternal retraction, the HU position appeared to prevent brachial plexus injury from posterior displacement of the shoulder, and may also minimize nerve stretch injury caused by first rib rotation. The HU position may also be beneficial in reducing ulnar nerve compression at the elbow which can occur with arm placement at the sides . Other investigators however, have noted that different arm positions made no difference in the incidence of brachial plexus injury during symmetric sternal retraction [1,6,7]. The HU position, as evidenced by this study, does not seem to afford complete protection from neural injury during IMA harvest, since patients positioned in this manner still had significant decreases in SSEP waveform amplitudes. However, the tendency toward a higher incidence of patients with neurologic symptoms in the AAS group compared to the HU group, coupled with the observation that three of seven patients in the AAS group had ulnar symptoms compared to none in the HU group, supports the possibility that ulnar nerve compression may produce some portion of brachial plexus symptoms observed with this positioning. When arm position was not considered, significant changes in SSEP waveforms consistent with nerve injury tended to be less common with the Pittman retractor compared to the Rultract retractor. These results are in keeping with previous work  which showed the Pittman retractor to cause less of a decrease in SSEP waveform amplitude when compared to the Rultract retractor during IMA harvest independent of arm position.
Other factors which could have contributed to brachial plexus injury were kept constant. Preexisting diabetic neuropathy might contribute to increased symptoms of brachial plexus injury, even after relatively minor, normally harmless insults to the nerve bundle . Diabetic patients were excluded from this study, as were any individuals with previous brachial plexus symptomatology. Advanced age is also a contributing factor to peripheral nerve injury during coronary bypass procedures [1,2]. There were no clinically significant differences in age between the groups studied here. Internal jugular cannulations, also linked to brachial plexus injury [7,17] were standardized with all placements occurring on the right side. Some of the patients in this study did complain postoperatively of right-sided symptoms. However, cannulations were performed with the patient awake but in a sedated state. If a large-bore needle were inserted into the brachial plexus, the pain would have been severe enough to alert the anesthesiologist of needle malposition. In fact, one of our patients did complain of pain during cannulation, which was immediately relieved after needle removal and redirection. This patient had no symptoms of brachial plexus injury after surgery. Brachial plexus damage has also been linked to the use of automated blood pressure cuffs . Here, the cuff was exclusively placed on the right side and appropriately positioned proximal to the ulnar groove where it was cycled rarely because of the use of an arterial catheter to monitor intraoperative blood pressure. Thus we believe no neural damage was attributable to the use of this monitor. Finally the duration of surgery has been correlated with brachial plexus injury suggesting the possibility of ischemic neuropathy . We could find no such correlation, since both groups had similar bypass, aortic crossclamp, and total anesthesia times.
In conclusion, our results demonstrate that asymmetric chest retraction during IMA harvest places the brachial plexus at significant risk of injury. The HU position offered no benefit in reducing brachial plexus injury associated with sternal retraction, except possibly when used in conjunction with the Pittman sternal retractor. HU positioning may also reduce the risk of ulnar nerve compression sometimes observed with AAS positioning. SSEP monitoring was capable of predicting damage to the brachial plexus, and may provide a means to determine the optimal positioning and retractor combination that will decrease plexus stress during IMA harvest for CABG.
1. Vander Salm TJ, Cereda JM, Cutler BS. Brachial plexus injury following median sternotomy. J Thorac Cardiovasc Surg 1980; 80:447-52.
2. Vander Salm TJ, Cutler BS, Okike ON. Brachial plexus injury following median sternotomy. Part II. J Thorac Cardiovasc Surg 1982;83:914-7.
3. Kirsh MM, Magee KR, Gago O, et al. Brachial plexus injury following median sternotomy incision. Ann Thorac Surg 1971;11:315-9.
4. Wey JM, Guinn GA. Ulnar nerve injury with open-heart surgery. Ann Thorac Surg 1985;39:358-60.
5. Tomlinson DL, Hirsch IA, Kodali SV, Slogoff S. Protecting the brachial plexus during median sternotomy. Thorac Cardiovasc Surg 1987;94:297-301
6. Roy RC, Stafford MA, Charlton JE. Nerve injury and musculoskeletal complaints after cardiac surgery: influence of internal mammary artery dissection and left arm position. Anesth Analg 1988;67:277-9.
7. Hanson MR, Breuer AC, Furlan AJ, Lederman RJ. Mechanism and frequency of brachial plexus injury in open-heart surgery: a prospective analysis. Ann Thorac Surg 1983;36:675-9.
8. Hudson DA, Boome R, Sanpera I. Brachial plexus injury after median sternotomy. J Hand Surg [Am] 1993;18:282-4.
9. Marganitt B, Shemesh Y, Golan M, et al. Subclinical brachial plexopathy following median sternotomy. Orthop Rev 1986;5:65-70.
10. Vahl CF, Carl I, Muller-Vahl H, Struck E. Brachial plexus injury after cardiac surgery. The role of internal mammary artery preparation: a prospective study on 1000 consecutive patients. J Thorac Cardiovasc Surg 1991;102:724-9.
11. Matthew EB, Vender JS, McGee JP. Upper extremity neuropathy after internal mammary artery dissection for coronary artery bypass. Anesth Rev 1987;14:42-5.
12. Hickey C, Gugino LD, Aglio LS, et al. Intraoperative somatosensory evoked potential monitoring predicts peripheral nerve injury during cardiac surgery. Anesthesiology 1993;78:29-35.
13. Jellish WS, Martucci J, Blakeman B, Hudson E. Somatosensory evoked potential monitoring of the brachial plexus to predict nerve injury during internal mammary artery harvest: intraoperative comparisons of the Rultract and Pittman sternal retractors. J Cardiothorac Vasc Anesth 1994;8:398-403.
14. John ER, Chabot RJ, Pricheps LS, et al. Real-time intraoperative monitoring during neurosurgical and neuroradiological procedures. J Clin Neurophysiol 1989;6:125-58.
15. York DH, Chabot RJ, Gaines RW. Response variability to somatosensory evoked potentials during scoliosis surgery. Spine 1987;12:864-76.
16. Graham JG, Pye IF, McQueen INF. Brachial plexus injury after median sternotomy. J Neurol Neursurg Psychiatr 1981;44:621-5.
17. Briscoe CE, Bushman JR, McDonald WI. Extensive neurologic damage after cannulation of the internal jugular vein. BMJ 1974;1(903):314.
18. Sy WP. Ulnar nerve palsy possibly related to use of automatically cycled blood pressure cuff. Anesth Analg 1981;60:687-8.
19. Sotaniemi KA. Brachial plexus lesion complicating sternotomy. J Neurol Neurosurg Psychiatry 1982;45:568.