Near-infrared spectroscopy can predict the onset of cerebral hyperperfusion syndrome after carotid endarterectomy.

BACKGROUND

Cerebral hyperperfusion syndrome (CHS) after carotid endarterectomy (CEA) is a potential life-threatening complication. Therefore, early identification and treatment of patients at risk is essential. CHS can be predicted by a doubling of postoperative transcranial Doppler (TCD)-derived mean middle cerebral artery blood velocity (V(mean)) compared to preoperative values. However, in approximately 15% of CEA patients, an adequate TCD signal cannot be obtained due to an insufficient temporal bone window. Moreover, the use of TCD requires specifically skilled personnel. An alternative and promising technique of noninvasive cerebral monitoring is relative frontal lobe oxygenation (rSO(2)) measured by near-infrared spectroscopy (NIRS), which offers on-line information about cerebral oxygenation without the need for specialized personnel. In this study, we assess whether NIRS and perioperative TCD are related to the onset CHS following CEA.

METHODS

Patients who underwent CEA under general anesthesia and had a sufficient TCD window were prospectively included. The V(mean) and rSO(2) measured before induction of anesthesia were compared to measurements performed in the first postoperative hour (ΔV(mean), ΔrSO(2), respectively). Logistic regression analysis was performed to determine the relationship between ΔV and ΔrSO(2) and the occurrence of CHS. Subsequently, receiver operating characteristic (ROC) curve analysis was used to determine the optimal cutoff values. Diagnostic values were shown as positive and negative predictive values (PPV and NPV).

RESULTS

In total, 151 patients were included, of which 7 patients developed CHS. The ΔV(mean) and ΔrSO(2) differed between CHS and non-CHS patients (median, interquartile range), i.e. 74% (67-103) versus 16% (-2 to 41), p = 0.001, and 7% (4-15) versus 1% (-6 to 7), p = 0.009, respectively. The mean arterial blood pressure did not change. Postoperative ΔV(mean) and ΔrSO(2) were significantly related to the occurrence of CHS [odds ratio (OR) 1.40 (95% CI 1.02-1.93) per 30% increase in V(mean) and OR 1.82 (95% CI 1.11-2.99) per 5% increase in rSO(2)]. ROC curve analysis showed an area under the curve of 0.88 (p = 0.001) for ΔV(mean) and an optimal cutoff value of 67% increase (PPV 38% and NPV 99%), and an area under the curve of 0.79 (p = 0.009) for ΔrSO(2) and an optimal cutoff value of 3% rSO(2) increase (PPV 11% and NPV 100%). The combination of both monitoring techniques provided a PPV of 58% and an NPV of 99%.

CONCLUSIONS

Both TCD and NIRS measurements can be used to safely identify patients not at risk of developing CHS. It appears that NIRS is a good alternative when a TCD signal cannot be obtained.