Near-infrared spectroscopy versus transcranial Doppler ultrasound for assessing dynamic cerebral autoregulation by transfer function analysis in sepsis

Dynamic cerebral autoregulation (dCA) is impaired with a prolonged cerebrovascular response time in critically ill patients with sepsis [1–4]. This could potentially expose the brain to intermittent hypoand hyperperfusion during acute changes in blood pressure, and thereby contribute to encephalopathy [5]. Near-infrared spectroscopy (NIRS) is a non-invasive real-time bedside technique, which has been highlighted for monitoring dCA in sepsis [6]. This is based on the notion that it provides an index of changes in cerebral cortical oxygenation (ScO2) which may be used as a proxy for cerebral blood flow (CBF) changes [7]. However, studies on both healthy volunteers and septic patients have reported that CBF changes assessed by NIRS may differ from those obtained by transcranial Doppler ultrasound (TCD) [8–11]. Here, we extend on previous findings in septic patients [1,11] by comparing NIRSto TCD-based estimates of dCA during spontaneous fluctuations in blood pressure. The Scientific Ethical Committee of Copenhagen and Frederiksberg Municipalities, Denmark, approved the study (file number H-A-2009-020 with amendments). Patients were included following informed consent from the next of kin, and from the patient’s general practitioner or the national health inspector. We obtained 15 20-min steady-state recordings of simulatenous invasive arterial blood pressure and NIRS and TCD signals in 10 mechanically ventilated patients admitted to a general tertiary ICU diagnosed with severe sepsis or septic shock within the past 72 h (Table I). Data on haemostasis and cerebral haemodynamic function, including a comparison of vasopressorinduced CBF changes assessed by NIRS and TCD on some of the patients, have previously been published [1,11,12]. We used a 2 MHz pulsed TCD (Ez-Dop and DopplerBox, Compumedics DWL GmbH, Singen, Germany) for the assessment of middle cerebral artery blood flow velocity (MCAv) with the Doppler probe secured over the transtemporal window, and an insonation depth of 45–60 mm. NIRS was performed by dual-wavelength technology (INVOS Cerebral Oximeter, Somanetics, Troy, MI, USA), with an optode attached to the patient’s forehead, ipsilateral to the TCD. Arterial blood pressure was recorded through an Invasive Blood Pressure Module M1006B (Philips Medical Systems, Böblingen, Germany). Data were sampled at 1 kHz using an analogue-to-digital converter (PowerLab 16/30TM, ADInstruments Ltd, Oxford, UK) interfaced with a personal computer. Recordings were re-sampled at 10 Hz, and transfer function analyis (TFA) was then performed by a 1200-point Fourier transformation, with a 60 s overlap between segments for Welch spectral estimation and multiplication by the Hanning window to minimise spectral leakage [13]. This was performed between arterial blood pressure and MCAv, and arterial blood pressure and ScO2, respectively. Data were analyzed in the very low (0.02–0.07 Hz), low (0.07–0.20 Hz) and high (0.20–0.30 Hz) frequency ranges. In each, frequency range, dCA was evaluated by calculating gain, phase, and coherence. Gain quantifies the effectiveness of dCA as a filter that dampens MAP-induced changes in CBF, while phase refers to the displacement of the CBF signal relative to the MAP signal, which reFects the response time of dCA, and coherence quantifies the