Secondary Logo

Journal Logo

Cerebral Oximetry Fails as a Monitor of Brain Perfusion in Cardiac Surgery: A Case Report

McAvoy, James MD*; Jaffe, Richard MD, PhD*; Brock-Utne, John MD, PhD*; López, Jaime MD; Brodt, Jessica MBBS

doi: 10.1213/XAA.0000000000000963
Case Reports
Free

Cerebral oximetry is commonly being advocated as a monitor for regional cerebral tissue oxygenation during cardiac surgery. We have increasing concern about the accuracy of this monitor, including the current systems entering the market, with new probes and algorithms. We present 2 cases where cerebral oximetry failed to accurately portray cerebral oxygenation. In the current form, cerebral oximetry may at best be an expensive tool without any benefit on outcomes. In addition, it may contribute to misleading and confusing clinical data.

From the Departments of *Anesthesiology, Perioperative, and Pain Medicine

Neurology

Anesthesiology, Perioperative, and Pain Medicine, Cardiac Anesthesia, Stanford University School of Medicine, Stanford, California.

Accepted for publication December 10, 2018.

Funding: None.

The authors declare no conflicts of interest.

Address correspondence to Jessica Brodt, MBBS, Department of Anesthesiology, Perioperative, and Pain Medicine, Stanford University School of Medicine, 300 Pasteur Dr, Room H3583, Stanford, CA 94305. Address e-mail to jbrodt@stanford.edu.

A recent study demonstrated the bias of cerebral oximetry under hypocapneic conditions.1 This finding is of particular interest given the routine use of deliberate hypocapnia in neurosurgery and alpha-stat management of adult patients during cardiopulmonary bypass (CPB). Currently, many of our cardiac and neurosurgical colleagues believe that cerebral oximeters can be used to monitor cerebral oxygenation and/or perfusion during their procedures. The underlying assumption is that higher (ie, >60%) cerebral oximetry values implies improved tissue oxygenation.

We have observed multiple issues with cerebral oximetry monitors and selected 2 representative cases where the displayed cerebral oximetry value did not reflect underlying cerebral function. Case 1 illustrates “normal” cerebral oximetry values despite intracerebral catastrophe. Case 2 shows a profound unilateral drop in cerebral oximetry despite normal neurologic outcome. In both cases, foreheads were prepped with alcohol and wiped dry before application of cerebral oximetry sensors over the frontoparietal skin on superior forehead, below the hairline.

These patients provided written consent under the Health Insurance Portability and Accountability Act.

Back to Top | Article Outline

CASE 1

An 81-year–old woman presented for redo sternotomy, triple-valve surgery, and total arch replacement for severe aortic stenosis with severe mitral and tricuspid regurgitation. She had previously undergone ascending aortic replacement for type A aortic dissection. Intraoperative neurophysiologic monitoring using electroencephalography, upper extremity somatosensory evoked potentials, and transcranial motor evoked potentials was used with oximetry (Fore-Sight, Casmed, Brandford, CT). After commencing CPB, electroencephalography and evoked potential signals were stable, displaying the changes expected with cooling, including global lateralized and generalized periodic discharges then burst suppression on electroencephalography, and symmetric and gradual decreases in amplitude with increased latency on evoked potentials. At 45 minutes before circulatory arrest (core temperature 28°C), there was an abrupt loss of bilateral somatosensory evoked potentials, followed by global slowing of the electroencephalography and loss of the transcranial motor evoked potentials without decreased cerebral oximetry. Cannulae, head position, pump parameters, and other physiologic variables were checked. Bilateral cerebral oximetry started to decline approximately 20 minutes after the loss in intraoperative neurophysiologic monitoring signals, as seen in Figure 1. When cerebral oximetry values began to decline, sensors were checked and found to be well adhered to skin. The area around the sensors was clean and dry. Further troubleshooting revealed that sensors were incorrectly assigned left and right (ie, left = “2” and right = “1”). To avoid monitor recalibration known to occur with cable disconnection, sensor replacement, or powering off and on, laterality was adjusted by renaming the correct sides on the cerebral oximetry base module so that “1” was left and “2” was right. No physical changes were made to cables, sensors, or connections. After correction of laterality on the base module, the right head cerebral oximetry abruptly increased to 72%, with no change in left head cerebral oximetry. The case progressed after discussion with the surgical team about the concerning intraoperative neurophysiologic monitoring changes and disparity seen on cerebral oximetry. Given the concerning changes in intraoperative neurophysiologic monitoring seen before circulatory arrest, increased flow (12 mL/kg/min instead of 10 mL/kg/min) was used during selective antegrade cerebral perfusion, with inflow temperature at 20°C.

Figure 1.

Figure 1.

Figure 2.

Figure 2.

There was no recovery in intraoperative neurophysiologic monitoring during rewarming and after weaning from CPB. The patient was transferred to the intensive care unit with satisfactory hemodynamics and hemostasis. A head computed tomography at <12 postoperative hours revealed right frontoparietal intraparenchymal hemorrhage, with bilateral intraventricular and subarachnoid blood (Figure 2). This was deemed nonsurvivable. She transitioned to comfort care and died later that day.

Back to Top | Article Outline

CASE 2

Figure 3.

Figure 3.

A 56-year–old woman with no known underlying carotid or cerebrovascular disease underwent orthotopic lung transplantation for interstitial fibrosis. Shortly after initiating CPB, the left cerebral oximetry decreased to 42% from a baseline of 77%, with the right maintained at 68%, close to a baseline of 61%. Sensors were well adhered, clean, and dry. Head position, cannulae, CPB, and physiologic parameters including Paco2 were checked and found to be within expected parameters. The left sensor was replaced, with an immediate increase to 68%, as shown in Figure 3. She was subsequently extubated in the intensive care unit and found to be neurologically intact.

Back to Top | Article Outline

DISCUSSION

These cases illustrate the limitations of cerebral oximeters in real clinical situations. Common known reasons for cerebral oximetry malfunction include poor skin contact, hair underneath sensor, or high light conditions (eg, surgical light shining on head). Cerebral oximeters sample cortical tissue at a depth of 2–3 cm, so any anatomy or pathology that increases the distance from skin to cortex will also cause erroneous cerebral oximetry readings (eg, frontal sinus pathology, skull hypertrophy). However, as shown in the cases described here, widely discordant values may be observed despite static conditions.

In case 1, the right frontoparietal hemorrhage (Figure 2) may have caused the initial decrease in right head cerebral oximetry, initially incorrectly attributed to the left side. When the laterality was corrected by renaming the sides on the base module (without changing probes, cables, or physical disconnection at any point), the cerebral oximetry value normalized. This should not occur in the setting of actual regional cerebral pathology. We contacted the regional representative for the device company, who facilitated a meeting among clinical faculty, company representatives, and regional leadership, as well as clinical engineering. The company took the module and cables used for the case for interrogation. Despite this interrogation and extensive discussions, their conclusion was that the changes we observed were “perplexing” and presumed to be an outlier incident. No further explanation was forthcoming about the apparent device failure.

In case 2, the leading hypothesis for sudden normalization of values was a unilateral faulty sensor, despite high baseline values and apparently normal function before the decrease. Other known causes of sensor malfunction were not detected during our troubleshooting. Normalization of unexplainable cerebral oximetry values after replacement of sensors is a pattern we have seen on multiple occasions in other cases. With the current generation of oximeters, there are no signal quality indices or any other indicators to discern whether values are low because of true pathology or whether it is a problem with the sensor or cable.

The industry push behind the use of these monitors is intense. In our institution alone, annual spending on cerebral oximetry is approximately $400,000. While these monitors may have utility in other clinical arenas, existing data show no compelling benefit of cerebral oximetry monitoring in cardiac surgery with regard to neurocognitive outcome, renal or myocardial injury, or overall mortality.2 Further, using cerebral oximetry as a targeted means of red cell transfusion demonstrated no benefit to patients; secondary outcomes in this study were similarly negative, including health care costs.3 These findings suggest that use of cerebral oximetry is insufficient to detect clinically meaningful cerebral hypoxia or that the mechanism of neurocognitive dysfunction after cardiac surgery is beyond the realm of detection by cerebral oximetry.

No monitor can—or is expected to—provide flawless data without the possibility of imperfect or artifactual values. Even with a near perfect monitor, human error in application or interpretation may lead to erroneous use of data. However, most monitors have some means of objectively assessing the quality of their measurements. This is not true for cerebral oximetry. For example, there is no way to determine the extent of signal contamination from superficial structures (eg, scalp, skull, meninges, cerebrospinal fluid) unrelated to cortical perfusion. Furthermore, even if regional cerebral oxygenation could be reliably measured, there is no proof that cortical hemoglobin oxygen saturation is consistently related to cortical tissue perfusion or viability. Any monitor that is relied on for clinical decisions must have a way of assessing the quality of its measurements or otherwise bears a substantial burden of proof that has not been attained by cerebral oximetry.

Here, we present 2 cases that illustrate the unreliability of cerebral oximetry, in the current form, for cardiac surgical procedures. While no monitor is perfect, data must be interpreted in the context of the global clinical picture, and our experience raises concern about the utility of these monitors to guide therapy during cardiac surgery.

Back to Top | Article Outline

DISCLOSURES

Name: James McAvoy, MD.

Contribution: This author helped prepare the manuscript, and write and revise the manuscript.

Name: Richard Jaffe, MD, PhD.

Contribution: This author helped write the manuscript.

Name: John Brock-Utne, MD, PhD.

Contribution: This author helped write the manuscript.

Name: Jaime López, MD.

Contribution: This author helped write the manuscript.

Name: Jessica Brodt, MBBS.

Contribution: This author helped prepare the manuscript, and write and revise the manuscript.

This manuscript was handled by: BobbieJean Sweitzer, MD, FACP.

Back to Top | Article Outline

REFERENCES

1. Schober A, Feiner JR, Bickler PE, Rollins MD. Effects of changes in arterial carbon dioxide and oxygen partial pressures on cerebral oximeter performance. Anesthesiology. 2018;128:97–108.
2. Serraino GF, Murphy GJ. Effects of cerebral near-infrared spectroscopy on the outcome of patients undergoing cardiac surgery: a systematic review of randomised trials. BMJ Open. 2017;7:e016613.
3. Rogers CA, Stoica S, Ellis L, et al. Randomized trial of near-infrared spectroscopy for personalized optimization of cerebral tissue oxygenation during cardiac surgery. Br J Anaesth. 2017;119:384–393.
Copyright © 2019 International Anesthesia Research Society