Continuous and Entirely Non-Invasive Method for Cerebrovascular Reactivity Assessment: Technique and Implications

Abstract

Purpose:

Continuous cerebrovascular reactivity assessment in traumatic brain injury (TBI) has been limited by the need for invasive monitoring of either cerebral physiology or arterial blood pressure (ABP). This restricts the application of continuous measures to the acute phase of care, typically in the intensive care unit. It remains unknown if ongoing impairment of cerebrovascular reactivity occurs in the subacute and long-term phase, and if it drives ongoing morbidity in TBI. We describe an entirely non-invasive method for continuous assessment of cerebrovascular reactivity.

Methods:

We describe the technique for entirely non-invasive continuous assessment of cerebrovascular reactivity utilizing near-infrared spectroscopy (NIRS) and robotic transcranial Doppler (rTCD) technology, with details provided for NIRS.

Results:

Recent advances in continuous high-frequency non-invasive ABP measurement, combined with NIRS or rTCD, can be employed to derive continuous and entirely non-invasive cerebrovascular reactivity metrics. Such non-invasive measures can be obtained during any aspect of patient care post-TBI, and even during outpatient follow-up, avoiding classical intermittent techniques and costly neuroimaging based metrics obtained only at specialized centers.

Conclusion:

This combination of technology and signal analytic techniques creates avenues for future investigation of the long-term consequences of cerebrovascular reactivity, integrating high-frequency non-invasive cerebral physiology, neuroimaging, proteomics and clinical phenotype at various stages post-injury.

Introduction:

Over the past decades, cerebrovascular reactivity monitoring has emerged as a strong associate with long-term outcome in traumatic brain injury (TBI).[1–5] Various retrospective and multi-center prospective studies have provided validating evidence to support the association between impaired cerebrovascular reactivity and both increased mortality and poor functional outcome after moderate and severe TBI.[6–10] Further to this, in the mild TBI literature, intermittent CO₂ based vascular reactivity assessments using advanced magnetic resonance imaging (MRI) techniques have also suggested a link between impaired CO₂ chemo-reactivity of the cerebral vessels and worse clinical phenotype after concussion/mild TBI.[11–13] Despite both of these literature bodies supporting the link between impaired cerebrovascular reactivity and patient outcome across the spectrum of TBI, the methods at which we measure and assess cerebrovascular reactivity/autoregulation carry important limitations.

In the setting of mild TBI, where invasive cerebral monitoring is not available, current assessments of autoregulatory capacity and cerebrovascular reactivity have relied in intermittent testing, focused on a dynamic physiologic challenge, such as: transient hyperemic response testing, thigh cuff deflation techniques, orthostatic hypotension testing and CO₂ manipulation.[12,14] Cerebrovascular response in this setting is typically measured pre- and post-intervention using either intermittent transcranial Doppler (TCD), which is classically user dependent and subject to issue with signal quality, or functional MRI techniques, which are costly and require transport to specialized facilities to be conducted. Furthermore, the symptomatic mild TBI patient may not tolerate various aspects of the above described testing, prolonged periods in a noisy MRI environment, or CO₂ based challenges.

For moderate and severe TBI, where admission to the intensive care unit (ICU) and use of invasive intracranial pressure (ICP) monitoring is common, the derivation of standard cerebrovascular reactivity indices, such as the pressure reactivity index (PRx – correlation between slow-wave vasogenic fluctuations in MAP and ICP) is possible.[1,2] Such continuous metrics have received support from international consensus conferences on multi-modal cranial monitoring in neurocritical care,[4,5] and have a growing literature body to support their application. The issue is that they rely on invasive cranial, and typically invasive arterial blood pressure (ABP), monitoring to obtain continuous uninterrupted signals for processing and index derivation.[1,2] This fact limits their application to the acute ICU phase of care, with no studies having been able to evaluate similar continuous metrics during the subacute or long-term phase post-TBI. It is unknown if ongoing impaired cerebrovascular reactivity in moderate/severe TBI, as assessed to such continuous metrics, is associated with long-term clinical phenotype.

Near infrared spectroscopy (NIRS) and robotic TCD (rTCD) are natural non-invasive devices for subacute and long-term phase application. NIRS, evaluating bilateral frontal cerebral oxygenated hemoglobin content, provides a surrogate measure for pulsatile cerebral blood volume (CBV). While, TCD and rTCD evaluates middle cerebral artery blood flow velocity (CBFV), providing a surrogate metric for cerebral blood flow (CBF). Using invasive ABP monitoring, NIRS and TCD/rTCD based metrics of cerebrovascular reactivity have been derived,[15–18] with NIRS metrics and TCD based systolic flow index (Sx) demonstrating some ability to discern the lower limit of autoregulation in experimental models.[19–21] Further, TCD-based Sx has been demonstrated to closely co-vary with standard invasive PRx measures,[16,22–25] suggesting it may provide a non-invasive surrogate assessment for PRx.

Unfortunately, these continuous NIRS and rTCD based measures have not been derived or applied in the subacute or long-term phase of care post-TBI, given the need for invasive ABP monitoring in the past to derive continuous mean arterial pressure (MAP).

In this technology report, we describe a completely non-invasive means for both NIRS and rTCD based cerebrovascular reactivity assessment, employing high-frequency non-invasive ABP and signal analytic techniques.

Technique Methodology:

Applicable Devices

Non-invasive surrogate assessments of pulsatile CBV or CBF can be conducted utilizing NIRS or TCD, respectively. We focused on the application of NIRS and non-invasive ABP for the purpose of demonstration of feasibility, though mention rTCD platforms within as an alternative to NIRS.

Commercially available NIRS platforms typically provide the ability to assess bilateral frontal oxygenated hemoglobin concentrations, displaying either raw values of oxygenated/deoxygenated hemoglobin levels, regional saturation metrics (regional oxygen saturation or rSO₂) or proprietary spatially resolved summary metrics such the total oxygen index (TOI) or total hemoglobin index (THI), in the case of the more advanced research oriented platforms such as the Hamamatsu NIRO devices (Hamamatsu Photonics Ltd, Japan, https://www.hamamatsu.com/jp/en/product/type/C10448/index.html). For this report, we focus on the bilateral application of the Covidien INVOS NIRS unit (INVOS 5100C, Covidien Canada, Medtronic, https://www.medtronic.com/covidien/en-us/products/cerebral-somatic-oximetry.html).

TCD platforms vary widely and have been classically limited in their long duration continuous application secondary to technique and user dependence. Recent description and application of rTCD for the continuous uninterrupted long-duration assessment of CBFV and cerebrovascular reactivity in critically ill TBI has been described.[25–27] Further, time-series analysis of rTCD based systolic parameters has demonstrated their close association with standard invasive PRx, suggesting their potential role as a surrogate measure for PRx.[24,25] Full description of these rTCD devices has been already conducted.[26] As such, these devices are only being mentioned here as an alternative to continuous NIRS, and will not be explored further.

Finally, continuous non-invasive high-frequency ABP can be obtained using finger-tip based methods employed in the Finapres devices. We highlight the newer Finapres NOVA device with the high-frequency Nanocore module (Finapres Medical Systems, Enschede, The Netherlands, http://www.finapres.com/home). This particular device allows for continuous non-invasive ABP assessment in full waveform at up to 250 Hz sampling frequency, with both high-frequency analoque and lower frequency serial output. Further, the brachial arterial waveform may be reconstructed when employing an upper arm cuff based calibration cycle. Figure 1 provides a pictorial overview of the NIRS and Finapres devices.

Figure 1:

Portable NIRS and Non-Invasive ABP Units

ABP = arterial blood pressure, NIRS = near infrared spectroscopy. Panel A – INVOS 5100C NIRS unit, monitor display and adhesive non-invasive frontal NIRS sensor (black arrow); Panel B – Finapres NOVA non-invasive continuous ABP monitor; Panel C – Finapres Nanocore wrist mounted ABP unit (dashed black arrow), finger-tip ABP cuff (white arrow), and height correction system (solid black arrow).

Set-Up

Figure 2 provides a representation of the set-up for continuous non-invasive NIRS based assessment of cerebrovascular reactivity. As with most commercially available NIRS devices, the application of bifrontal sensors is simple, employing bilateral disposable adhesive pads in the case of the displayed INVOS 5100C device. The display for the INVOS 5100C can be seen with bilateral simulatenous recording of rSO₂. The Finapres NOVA Nanocore non-invasive continuous ABP system is displayed, with the wrist mounted Nanocore module, finger-tip cuff and height correction device. The full digital touch-screen display for the Finapres NOVA can be seen with both the finger and reconstructed brachial arterial waveforms. Finally the analogue signal acquisition components can be seen exiting the Finapres device and entering the laptop.

Figure 2:

NIRS and Non-Invasive Continuous ABP Real-Time Data Collection and Set-Up

ABP = arterial blood pressure, NIRS = near infrared spectroscopy. Panel A – INVOS 5100C bilateral frontal NIRS based regional cerebral oxygenation; Panel B – Finapres NOVA with Nanocore wrist mounted module collecting full waveform non-invasive continuous ABP from fingertip cuff, with height correction. Display demonstrates both finger ABP and reconstructed brachial artery waveforms; Panel C – analogue-digital signal converter pulling high-frequency full waveforms through custom 3.5mm jack-to-BNC, then feeding via USB into computer; Panel D – example of full NIRS and Finapres set-up when applied to a human.

Signal Acquisition for Demonstration

High frequency digital data is acquired using a bedside laptop. Any data acquisition software can be utilized, though for the purposes of this demonstration we employed intensive care monitoring plus (ICM+) (Cambridge Enterprise Ltd, Cambridge, UK, http://icmplus.neurosurg.cam.ac.uk). Data from the INVOS 5100C platforms can be exported in serial ASCII protocol format. While the rTCD systems can export using either serial or analogue output methodologies. The Finapres NOVA system can also export either using serial or analogue output, for the purpose of this demonstration of technique we employed high-frequency analogue export to obtain full non-invasive ABP waveforms. Data acquisition for analogue signals was conducted using external analogue-digital signal converters (ADC) (DT9804, Data Translation, Marlboro, MA), employing custom 3.5 mm jack-to-BNC cables to export data from the Finapres to the laptop. All physiologic signals are recorded and locked in time-series format, allowing for signal analysis and derivation of continuous metrics of cerebrovascular reactivity. Figure 2 displays the ADC-laptop set-up.

Example of Recorded Physiology

Figure 3 provides an example of the raw record physiology for the continuous non-invasive ABP and bilateral frontal regional oxygen saturation (rSO₂) from the INVOS 5100C device. Data was recorded continuously for 1 hour, with the finger-cuff switched at 30 minutes on the Finapres device. Of note is the high quality, full waveform nature of the non-invasive ABP. Signal artifact was manually removed in ICM+.

Figure 3:

Raw Physiologic Data Recorded in Locked Time-Series Format from NIRS and Finapres nABP

BABP = reconstructed non-invasive brachial arterial blood pressure, FABP = non-invasive finger-tip arterial blood pressure, mmHg = millimeters of Mercury, nABP = non-invasive arterial blood pressure, NIRS = near infrared spectroscopy, rSO₂_L = regional cerebral oxygen saturation (Left), rSO₂_R = regional cerebral oxygen saturation (Right), % = percent. Figure displays raw real-time physiologic data from Finapres non-invasive arterial blood pressure monitoring and bilateral NIRS. Lower panels display full arterial waveforms from finger-tip and the reconstructed brachial artery waveform. Sampling frequency for Finapres system at 100Hz in this example.

Figure 4 provides an example from ICM+ of continuously derived bilateral cerebrovascular reactivity using NIRS. The cerebral oxygenation index (COx), was derived using the moving correlation between slow-wave vasogenic fluctuations in rSO₂ and MAP. These slow-wave values were obtained by applying a 10-second moving average filter, decimating the signal to the frequency range associated with cerebral autoregulation (ie. 0.05 to 0.005 Hz). These bilateral metrics displayed in Figure 4 highlight the feasibility of an entirely non-invasive and continuous cerebrovascular reactivity assessment, with bilateral hemispheric measurements. Given the set-up highlighted in Figure 2, it can be see how this method can be employed during the subacute hospital phase, and outpatient clinical settings.

Figure 4:

Example of Continuously Derived Bilateral Non-Invasive Cerebrovascular Reactivity Indices

a.u. = arbitrary units, COx_L = cerebral oxygenation index (correlation between rSO₂_L and MAP), COx_R = cerebral oxygenation index (correlation between rSO₂_R and MAP), MAP = mean arterial pressure, rSO₂_L = regional cerebral oxygen saturations (Left), rSO₂_R = regional cerebral oxygen saturations (Right), % = percent. Figure displays 10-second moving average data, with bilateral COx indices derived using the moving correlation between slow-waves in rSO₂ and MAP, updated every 10-seconds for this case example.

Limitations of Technique/Technology

The above demonstration of the derivation of non-invasive cerebrovascular reactivity metrics is interesting, though there are some limitations. The biggest is cost. The non-invasive ABP devices are costly, as rTCD and some NIRS platforms. Disposables are non-existance for rTCD and the Finapres devices, but are something to consider for NIRS platforms, with specific platforms having much higher costs for the patient NIRS disposables. Second, the Finapres system does require intermittent alternating of finger or cuffs for extended duration recordings. This does produce interrupted waveforms during this process, though they are short in duration. Third, this overall technique, whether using NIRS or rTCD, does require some biomedical engineering expertise for set-up, signal acquisition and data processing.

Finally, regarding current NIRS technology, there does exist some limitations in obtaining pure cerebral signals. Various devices exist for continuous cerebral oximetry, including the INVOS unit discussed in our techniques. Each device varies in its ability to provide true pure intracranial assessments of cerebral oxygenation saturations, or oxy- and deoxy-hemoglobin concentrations (depending on the device). Various proprietary algorithms, designed to exclude contamination from the extracranial scalp circulation, are employed by the manufacturers, and are device specific. These devices, with their algorithms, are commonly referred to as ‘spatially resolved’ NIRS units, with the main focus of the continuously monitored signal being intracranial in nature. With that said, these devices are still subject to varying degrees of signal contamination from extra-cranial sources.[28] Further to this, NIRS devices are providing information regarding regional oxygen saturation, or oxy-hemoglobin, status. As such, this value is subject to many factors influencing oxygenation, from oxygen uptake in the lungs, to end organ delivery in the brain. Therefore, proper interpretation of NIRS needs to occur in context of the patient’s cardiovascular, respiratory and hemoglobin status. Finally, the region of brain being monitored requires some consideration, as the signals obtained from those with significant scalp injury, subgaleal hematomas, or cerebral contusions are difficult to interpret and should be considered spurious in nature given current technologic limitations.

As such, given situations where NIRS may be limited or difficult to interpret, rTCD platforms may have the opportunity to shine. Though quite expensive and requiring the need for some specialized training, they can circumvent some of the above NIRS related limitations, and provide extended duration uninterrupted intra-cranial CBFV signals,[25–27] which can be utilized in concert with nABP via devices such as the Finapres. As such, the specific technique/devices utilized to obtain continuous and entirely non-invasive cerebrovascular reactivity monitoring needs to be tailor to the specific patient.

Prior to widespread adoption and application of the above described technique, there will need to be validation studies, conducted in health and diseased states, evaluating different NIRS and rTCD platforms, using the individual patient as their own control.[28] Such work may shed light on which NIRS or rTCD platform performs superiorly, carry the least amount of noise and signal contamination. Such work will be the focus of ongoing studies from our group, in conjunction with our national and international collaborators.

Implications and Future Applications:

Use of the described techniques carries important implications for assessment of patients with various neuropathological states, including TBI, during the subacute and long-term phases of care. To date, our assessment of cerebrovascular reactivity, after the removal of invasive cerebral monitoring devices such as ICP, has been limited to intermittent techniques based on TCD or neuroimaging, which are quite dissimilar to standard invasive monitoring employed in the ICU. Employing NIRS or rTCD, one is able to obtain non-invasive surrogate measures of pulsatile CBV or CBF. Integrating these with continuous non-invasive high-frequency ABP measures, such as those obtained through the Finpres device described, signal processing techniques may be applied to the acquired data to derive continuous metrics of cerebrovascular reactivity. Furthermore, hemispheric based measures may be derived, allowing one to comment on symmetry of cerebrovascular reactivity[29] in an entirely non-invasive way.

Such TCD and NIRS metrics have been derived in the past using invasive ABP to derive MAP, and have been shown to measure aspects of autoregulation experimentally. Further, these NIRS and rTCD measures have been shown to closely co-vary with the standard invasive PRx metrics, and may even be able to be used as surrogates. The application of newer continuous non-invasive ABP devices allows the extrapolation of such NIRS and TCD cerebrovascular reactivity measures to phases of care where invasive monitoring in unavailable or inappropriate. Such settings include the subacute hospital or rehabilitation phases of care. Similarly, outpatient clinical visits can now integrate this type of non-invasive cerebrovascular reactivity assessments, without the need to transfer to other centers for costly neuroimaging. The important aspect of these non-invasive continuous metrics is that they are similar to the continuous invasive metrics currently being employed in the ICU for moderate and severe TBI,[16,22,23] allowing for direct comparison between acute and subacute/long-term phase measures. The same cannot be said for neuroimaging, or other intermittent dynamic testing techniques. We provide two patient population examples of where this type of monitoring may be of benefit: aneurysmal subarachnoid hemorrhage and TBI.

First, this non-invasive methodology may play a key role is in the early detection of cerebral vasospasm after aneurysmal subarachnoid hemorrhage.[30–33] Emerging literature points to a direct association between impaired cerebrovascular reactivity and the development of symptomatic cerebral vasospasm after aneurysmal subarachnoid hemorrhage.[30] To date, this type of continuous cerebrovascular reactivity monitoring has been conducted using invasive ICP devices (such as ventricular drains or parenchymal monitors).[33] However, many aneurysmal subarachnoid hemorrhage patients do not require such invasive monitors on presentation, though are at significant risk for the development of cerebral vasospasm, leading the cerebral infarcts and significant morbidity and mortality. Currently, for such a population we are left with intermittent bedside TCD or neuroimaging techniques to detect altered cerebral vessel caliber.[34,35] These techniques suffer from both user dependent signal acquisition and poor temporal resolution. As such, continuous non-invasive cerebrovascular reactivity assessments, via NIRS or rTCD, may prove to provide continuous bedside early warning of impending symptomatic cerebral vasospasm, allowing for early intervention to reduce the risk of cerebral ischemia and infarction. Similarly, this monitoring may provide continuous insight into the effectiveness of specific therapies directed at cerebral vasospasm, providing a physiologic measure, in addition to clinical phenotype, to titrate therapeutic measures to.

Second, applying this constellation of technology for the continuous non-invasive assessment of cerebrovascular reactivity in TBI allows for exploration into the association between acute phase impairment and: long-term cerebrovascular reactivity status, protein and cerebrospinal fluid biomarkers of injury and neuroinflammation, and long-term clinical phenotype. It is becoming increasingly clear that impaired cerebrovascular reactivity is a strong predictor of TBI patient outcome at 6 months, independent of ICP.[1,8,9,36] Furthermore, in the acute phase of TBI care, most patients spend upwards of 50% of a given day with impaired cerebrovascular reactivity,[37] with current treatment provided for TBI failing to target a patient’s cerebrovascular reactivity status.[8,37] Current continuous monitoring of cerebrovascular reactivity focuses on ICP derived metrics,[1] which are reliant on invasive ICP monitoring, and only available when the monitoring device is in situ. As such, we lose our ability to monitor cerebrovascular reactivity once the ICP monitor is removed, while the patient likely suffers ongoing dysfunction. Such dysfunction is likely associated with ongoing fluctuations in clinical phenotype, long after ICP control issues have been treated. It is well known from the mild TBI literature that ongoing cerebrovascular CO₂ reactivity impairment is seen in those patients still suffering from persistent post-concussion symptomatology.[11–13] Such persistent symptoms are now the focus of return-to-play and return-to-learn initiatives.[38,39] With the proposed non-invasive continuous assessment of cerebrovascular reactivity, we can in theory replace the need for expensive and extended duration functional neuroimaging, and provides bedside or clinic-based assessments of cerebrovascular status. This type of monitoring can be correlated with clinical phenotypes, across the spectrum of TBI, to provide information during states of neurological fluctuations or persistent symptomatology in the long-term. Furthermore, as we uncover the specific mechanisms involved in impaired cerebrovascular reactivity, and develop therapeutics, this type of monitoring will be crucial for assessment treatment response.

Further studies are planned in our group, and collaborative networks, to integrate this technology and technique for the assessment of healthy control populations and TBI populations, across the spectrum of TBI severity and through various stages of their care, correlating with imaging, proteomic and clinical phenotypes. This work will also integrate different NIRS and rTCD platforms, as mentioned previously, in order to determine how different devices perform, given know inter-device variance in signal acquisition.[28] Such work is positioned to shed light on ongoing cerebral physiologic dysfunction in TBI, and open doors for future investigation of drivers of cerebrovascular impairment after TBI, leading to therapeutic targets for prevention and treatment.

Conclusions:

The application of NIRS or rTCD with continuous non-invasive ABP monitoring, provides the ability to derive completely non-invasive continuous metrics of cerebrovascular reactivity. This technique may be applied in the subacute hospital stay or outpatient clinical phase of patient care for those suffering from TBI, or other neuropathological states. Such metrics are closer surrogate measures of standard invasive continuous metrics employed during the ICU phase of care, compared to classical intermittent or MRI based techniques, and open the door for a wide array of investigations into the long-term consequence of impaired cerebrovascular reactivity.