Discussion
Despite its benefits in abbreviating side effects of deep hypothermia and longer cardiopulmonary bypass periods, ASCP with moderate hypothermia bears risks of warm ischemia for the abdominal viscera and especially spinal cord during lower body circulatory arrest if prolonged operations times are required or technical challenges occur unexpectedly. Spinal cord injury and ensuing paraplegia after aortic arch surgery is a rare but devastating complication for the patients. Spinal cord perfusion is maintained by segmental arteries from the thoracic and abdominal aorta and increased with the contribution of left subclavian, intercostal, lumbar, hypogastric arteries and the local paraspinous arterial tree embedded in the paraspinous muscles according to the current collateral network concept 6,11,12. As ASCP eliminates the segmental and hypogastric artery contribution to the spinal cord blood supply it is suggested that nearly two thirds of total flow is lost leaving the vertebral and intercostal arteries only to perfuse the cord 6,11-13. Routine addition of distal aortic perfusion to overcome this problem may complicate the surgical field and lengthen the procedure with its inherent problems related to retrograde femoral artery perfusion 6. On the other hand, additional subclavian perfusion may cause significant back bleeding from the intercostal arteries to the descending aorta causing a steal phenomenon and insufficient perfusion of the lumbar segments although blood to the cervical and mid-thoracic segments can be provided14. Therefore, when additional distal perfusion is not used, duration of ischemic period and the degree of hypothermia become the critical issues for spinal cord protection during arch surgery.
There is no consensus as to the safe duration of ASCP to rule out spinal cord ischemic injury during arch surgery with lower body circulatory arrest. As most reports are based on experimental studies concerns about safe period remain 1,15-17. Spinal cord injury was inevitable after 120 minutes and probable after 90 minutes in an experimental study at 28°C. In the clinical setting, when temperatures between 25-28°C are considered, Kamiya et al reported that LBCA duration up to 60 minutes was safe but beyond that limit paraplegia rate increases 18. The rate was 0% with deep hypothermia (20°C to 24.9°C) with bilateral ASCP but when patients with lower-body circulatory arrest greater than 60 minutes were further evaluated, overall paraplegia rate rose to 2.1% (8/377). They concluded that moderately hypothermic LBCA can be safely applied up to 60 minutes. Beyond that limit they question the safety. Paraplegia rates are usually 5 to 8% in contemporary series but rates as high as 18.2% have been reported with moderate hypothermia (25-28°C) 5,18. Zierer et al reported 0.3% paraplegia rate in 1002 patients operated at 26-34°C and mostly unilateral ASCP (67.2%). All patients with paraplegia had aortic dissection with a wide range of ASCP period (24, 41 and 127 minutes) 19. In another large series of patients operated with moderate hypothermia (20-24°C) occurrence of paraplegia was (0.5%) attributed to the creation of elephant trunk20. Elephant trunk formation causes a delay for delayed reperfusion injury or inadequate perfusion. Mean ASCP duration was relatively short (25.13 ± 19.02 minutes) in the present study (range 10-90 min) with 3 patients’ duration exceeding 60 minutes (63, 64 and 90 minutes). The only patient in the present study with postoperative paraparesis (%3.3) had previous chronic aortic dissection, 90 minutes of ASCP and elephant trunk formation consistent with the above mentioned studies.
NIRS is effectively used for monitoring cerebral oxygen saturation during ASCP in aortic arch surgery and has also been shown to be feasible for spinal cord ischemia detection in procedures involving thoracoabdominal aorta 4. As it reflects mixed tissue oxygenation -both arterial and venous blood hemoglobin oxygenation- rather than absolute values, usually a decrease of 20–30% from the reference level suggests a possible risk of ischemia. For cerebral NIRS monitoring during aortic surgery, a relative rSO2decrease to 64–80% of baseline (corresponding to 36-20% decrease of baseline) is suggested to be closely associated with neurologic events by the Japanese Society of Cardiovascular Anesthesiologists21. As there is no such predetermined critical NIRS value in the literature for spinal cord injury to occur we evaluated both 20 and 36% reductions in rSO2 values in our study group. More than half of the patients experienced more than %20 drop in T5 and T10 levels and only one of them had postoperative paraparesis. Less patients had more than %36 drop in their baseline NIRS values during ASCP as can be expected but none of them suffered from paraplegia. The only patient with paraparesis had NIRS reductions in T5 level by 31.5% and T10 level by 34.7% in T10 which returned to 22.2% drop in baseline after bilateral ASCP was instituted at the 33rd minute. However, his ASCP period was long enough to cause paraparesis (90 minutes). On the other hand, the only patient with more than 36% reduction in both T5 T10 values had only 17 minutes of ASCP with no postoperative complication. It is clear that the time period during which NIRS values remains low is very important. But we could not record instantaneous time related changes in NIRS values during the operation which is one of the limitations of the present study.
As to the optimal NIRS optode placement location for spinal cord, which is surrounded by vertebral bodies, supporters of positioning over the bilateral paraspinal muscles argue that these muscles supplied by the longitudinal collateral network indirectly reflects spinal cord perfusion 22. On the other hand, it is claimed that placing optodes on both sides of the vertebral column over the paraspinal muscles interfere with each other if the light from the contralateral side is received 23. Thinking that poorly vascularized adipose tissue overlying the muscles especially in obese patients may obscure light transmission if it is thick enough to exceed the penetration depth of NIRS light (12 to 20 mm). Therefore, we chose to place the optodes over the vertebral bodies distal to the spinal processes.
The level to which the NIRS optodes should be placed is another issue. When proximal aorta is clamped 3 cm below the subclavian artery leaving both subclavian arteries intact, high thoracic region (T4-6) levels do not show major fluctuations in NIRS value owing to extensive collateralization in this area 5. Mid thoracic (T7-9) and low thoracic (T10-12) values on the other hand drop significantly with high lumbar (T13-L2) and low lumbar (L2-5) values being the most prominent. As there is not such extensive collateralization below T8-9 thoracic level, mid-thoracic to lumbar region monitoring is usually preferred in thoracoabdominal interventions and regarded as a baseline reference to the more caudally placed optodes 1,5. In aortic arch surgery however, where associated coronary artery bypass grafting and harvesting of the left internal thoracic artery is a common procedure, inclusion of high thoracic levels for spinal cord NIRS monitoring should be undertaken as one major collateral is sacrificed (LIMA use in 23.3% of the cases in the present series). A fairly good positive correlation was observed between T5 and T10 levels in the present study when thoracic NIRS values fell significantly during lower body ischemic period (ASCP). In a pioneering study conducted by Kinoshita et al in patients undergoing arch replacement optodes of NIRS were placed at the level of 3rd and 10th thoracic vertebra over the paraspinous muscles and they have observed that regional oxygen saturations fell significantly with the start of ASCP 8. While T10 values continued to decrease to about 20% of baseline T3 NIRS levels remained relatively stable unlike the present series. They a had similar rate of CABG (17% and 23.3%) and most probably a similar rate of internal thoracic artery harvesting with our series but we think that additional perfusion of the left carotid and left subclavian artery might have accounted for the higher T3 NIRS values in their practice. They reported that with the advent of lower body circulation via a side branch from the graft the difference between 4 NIRS optodes (right and left forehead, T3 and T10) disappeared. They recovered 30 minutes after resumption of CPB similar to the present study. They did not observe any paraplegia in their 18 patients and concluded that cooling was the most important means of protecting the spinal cord from injury (core temperature about 25°). They had a higher mean lower body circulatory arrest compared to our study (49±14 vs 25.13 ± 19.02 minutes) but when LBCA times are compared in both studies they did not have any patient with more than 90 minutes of LBCA duration (ranges between 24 and 72 minutes). The only patient with postoperative paraparesis in our study had 90 minutes of LBCA. Therefore, we can conclude that when LBCA durations are expected to approach 90 minutes, additional lower body perfusion should be implemented as was also shown in animal studies1.
One important result of the present study is moderately negative correlation between the lactate levels in blood obtained from descending thoracic aorta and T10 NIRS values during the ASCP period. Lactate levels from the radial artery and descending aorta were moderately correlated (r=0.655) but they were significantly higher in descending aorta probably due to the retrograde blood flow to descending aorta from the visceral, thoracic and lumbar arteries rich in lactate. We think that lactate level measurement during open distal anastomoses accompanied with T10 NIRS monitoring can be a useful adjunct during arch operations with long ASCP periods. However, studies with larger volumes may help to determine a threshold limit to be used as an alarming sign for spinal cord ischemia.
Regarding s100β levels, in patients with spinal cord injury after thoracoabdominal operations Kunihara et al. observed that serum levels reached peak levels just after the operation and patients with spinal cord injury had a second peak at 24 hours postoperatively (1.3 ± 1.4 μg/L) 24. We do not have a baseline s100β value preoperatively, but the values in 6th postoperative hour fell significantly compared to values at the end of ASCP. Our values are somewhat higher but serious neurological complication occurred in one patient whose s100β levels were 0.33 ng/ml and 1.929 ng/ml respectively and one patient is not enough to make a reliable statistical evaluation.
One of the most important drawbacks of the present study is failure to monitor lumbar spine NIRS values. We could have observed significant fall in lumbar NIRS values before any reduction occurred in the thoracic level. Another shortcoming is absence of basal s100β level determination before any procedure took place. Patients with acute aortic dissection was excluded but there were 2 patients with chronic aortic dissection with previous intervention to the ascending aorta and aortic valve. Instantaneous display of NIRS during the operation was available however we could not record them. This study was conducted in a small patient group. Large scale studies are needed to arrive at a better conclusion regarding alarming signs for spinal cord injury during aortic arch surgery.