Low b-value diffusion weighted imaging is promising in the diagnosis of brain death and hypoxic-ischemic injury secondary to cardiopulmonary arrest

Evaluation of brain perfusion is critically important in patients after cardiorespiratory arrest. These patients suffer an initial hypoxic ischemic injury which can lead to cytotoxic edema as well as global hypoperfusion and brain death (BD). In these patients, ancillary tests can evaluate perfusion status when the clinical examination is limited [1]. Over 40% of subjects with BD are viable organ donors; however, only 10% of subjects with cardiac arrest go on to BD. Because delayed diagnosis of BD has been found to contribute to the overall shortage of viable organ transplants, some have argued for earlier use of ancillary testing in subjects where examination in this critical diagnosis is obscured by central nervous system (CNS) depressant medication [2, 3].

Introduction of targeted temperature management has significantly improved the clinical course in comatose cardiorespiratory arrest patients, with some patients being discharged with minimal brain damage. However, neurologic assessment during this time has remained a critical barrier since sedation and muscle paralysis from hypothermia obscure the clinical examination in the first 24–48 h, a crucial time for these patients [4, 5]. It is suspected that sedation-obscured clinical examination may contribute to withdrawal of life-sustaining therapy in up to 20% of patients during this time who may have otherwise had complete neurologic recovery [4, 5]. Conversely, CNS depressant medications have also been linked to a delay in prognosis of patients with BD, which has negative implications for organ transplantation viability and has led to increased mortality in cardiac transplant recipients [2]. These two extremes demonstrate the clinical dilemma in these unresponsive patients under hypothermic therapy or CNS depressants: the discrimination between subjects sustaining global injury with maintained cerebral blood flow, and subjects with cessation of blood flow and irreversible loss of function defined as BD [6].

Single photon emission computed tomography (SPECT) nuclear scintigraphy is an accepted ancillary test for BD [7‐12]. Transcranial Doppler has also been used as a prognostic tool, showing poor outcomes for patients with hypoperfusion, and good prediction of survivability in patients with normal perfusion [13, 14]. Computed tomography angiography (CTA) has been found to be comparable to ancillary tests, but lacks sensitivity [1]. In magnetic resonance imaging (MRI), the presence of bilateral transcerebral and cortical vein signs on susceptibility-weighted imaging (SWI) are helpful indicators in patients with both BD and hypoxic ischemic encephalopathy (HIE), although they are not specific [7, 15]. MRI in BD also consistently shows loss of vascular flow-voids on T2 spin echo and flow-related enhancement on magnetic resonance angiography (MRA), and diffuse cerebral edema with sulcal effacement [7]. Arterial spin labeling (ASL) has been validated as an effective noncontrast method for demonstrating perfusion deficits by demonstrating global hypoperfusion in BD with diffusely impaired cerebral blood flow on visual and quantitative analysis [16, 17]. Diffusion weighted imaging (DWI) shows promise, and apparent diffusion coefficient (ADC) values obtained at high gradient strengths (b750–2000) have been shown to be helpful prognostic indicators in patients with HIE, with decreased ADC values predicting poorer outcomes [18‐20]. Still, current use of high b-value DWI does not quantify perfusion and cannot be used to discriminate between the cellular injury of HIE and hypoperfusion related to BD.

DWI measures hydrogen motion in tissues. In addition to intra-/extracellular water self-diffusion, DWI is also affected by bulk water movement (cerebrospinal fluid (CSF)) and intravascular motion [21‐23]. While high b-value DWI measures molecular self-diffusion, low b-values are more sensitive to blood flow [21, 23‐25]. This is the basis of intravoxel incoherent motion (IVIM), a noncontrast sequence allowing derivation of perfusion (vascular) and diffusion (extravascular) fractions and is quantified by ADC [23] which has been used to evaluate acute stroke, dementia, as well as brain and head and neck masses [26‐32]. Though IVIM has many advantages, including its independence from arterial timing and its ability to quantify perfusion without contrast, postprocessing is time consuming and imaging times can be long since multiple b-values are necessary for processing [26, 33]. The current view is that b-value gradients between 0 and 200 s/mm² primarily measure perfusion, whereas gradient strengths greater than 200 s/mm² primarily measure molecular diffusion [22, 34].

We hypothesized that low b-value DWI could also be a helpful diagnostic tool in detecting global hypoperfusion. Specifically, our goal was to determine if low b-value DWI could separate the cellular injury of HIE from the global hypoperfusion of BD. To test this hypothesis, we evaluated patients after cardiorespiratory arrest undergoing MRI with low b-value DWI.

This retrospective study was conducted under an Institutional Review Board approved protocol and informed consent was waived. Investigators were compliant with the Health Insurance Portability and Accountability Act and Good Clinical Practice guidelines.

Evaluation of patients after cardiorespiratory arrest is a clinical challenge. We found that using a DWI sequence acquired at a single low b-value on a 1.5 T magnet was a promising technique for discrimination of BD from HIE and normal perfusion.

Results reinforce the concept that low b-values primarily measure brain perfusion properties, rather than diffusion, as has been previously established in the setting of acute stroke (Fig. 3) [26, 29, 30, 32, 36]. Separation of these properties is best seen in the HIE patients who exhibited profound regions of diffusion restriction on b1000–2000 images indicating cellular injury but showed no perfusion restriction on the b50 sequence, in fact showing elevated ADC values in both gray and white matter regions (Fig. 1).

ASL has successfully confirmed BD both quantitatively and qualitatively; however, the necessity for postprocessing and reliance on proper inflow tagging make it less user friendly and potentially less reliable [16, 24, 37]. Because the b50 sequence is “local” in the sense that it provides perfusion information without reliance on these distant processes, it does not suffer from the same potential inaccuracies.

The b50 DWI sequence also has benefits compared with usage of conventional MRI sequences, as it can give quantitative perfusion information compared with the qualitative evaluation of herniation and flow voids. This was reinforced in our results where the b50 ADC values in gray matter corresponded better with clinical diagnosis of BD than the presence of flow voids (which were present at the skull base in 2/9 BD subjects). This single low b-value sequence also has benefits in comparison with IVIM, not least of which are its rapid acquisition time, no need for offline postprocessing, and no specific software as it is a simple modification of a standard three-direction diffusion-weighted sequence. While b50 includes both cellular and vascular information, unlike IVIM which separates these components, it is heavily weighted towards perfusion and therefore able to primarily demonstrate perfusion changes. Visually, this sequence had a distinct appearance in BD with diffuse hypointensity from global perfusion restriction (Fig. 1).

Findings suggest that HIE patients have increased perfusion compared with normal controls, in direct contrast to the global hypoperfusion findings in BD. This is in agreement with prior studies demonstrating hyperperfusion with HIE using both ASL and contrast-based perfusion, reflecting the pathophysiologic course of whole-brain insults where initially there is overcompensation of flow to injured areas [38‐43].

A single outlier was present in the BD population which demonstrated more elevated white matter b50 ADC values (in agreement with HIE/normal values) than the rest of the BD subjects (Fig. 2b). This subject also had mildly elevated gray matter values in comparison with the other BD patients; however, these values were still distinctly lower than those in the HIE/normal subjects (Fig. 2a). This subject demonstrated bilateral parasellar carotid flow voids which extended to the bilateral M3 segments on MRI, although all other MRI findings supported BD (diffuse edema, herniation, and diffusion changes consistent with anoxia). Clinically, this subject had a much shorter timeframe from ROSC to MRI than the other BD subjects, being imaged within 2 h of ROSC as compared with 1–3 days in all other BD patients. The small interval between ROSC and MRI, and the presence of flow voids, support the possibility that this subject had a small residual amount of perfusion at the time of scanning which accounted for the higher b50 ADC values, though no ancillary testing was performed to confirm this. This supports that ADC values in b50 DWI are primarily affected by perfusion, as this patient demonstrated higher values than the BD patients who demonstrated loss of flow voids at the skull base (Fig. 4). This may give insight into the pattern of perfusion loss in BD, with gray matter regions losing perfusion before white matter, although this is only speculative in such a small study population. These findings support that low b50 ADC values in gray matter are more predictive of clinical BD than those in the white matter, and are more predictive of BD than the absence of flow voids at the skull base. A larger study population with confirmatory ancillary testing is necessary to further delineate the optimal b50 ADC threshold for this diagnosis.

Study limitations include vulnerability of echo-planar DWI to motion, chemical shift, susceptibility, and N/2 ghost artifacts, and the use of two different 1.5 T magnets in the study population, all of which could cause inaccuracies in ADC measurements. Additionally, this study is limited by small power and retrospective case-control comparisons. Prospective studies involving larger numbers of BD patients across site, scanner, and sequence architecture are warranted to further establish this technique.

Because BD is primarily a clinical diagnosis, and imaging is only used when the clinical examination cannot be fully obtained or results are in question, the majority of our BD subjects did not have correlative perfusion imaging (7/9). While this is a weakness in terms of validating this particular sequence, low b-value imaging itself has been previously validated as a reliable noncontrast perfusion technique [23‐26]. Our study demonstrating wide discrimination of mean ADC values with lack of overlap in the BD group in comparison with the perfused normal and HIE groups in gray matter serves to further support this technique. Although the clinical examination remains the gold standard for diagnosis of BD, this low b-value technique may be a promising supportive tool and deserves further study.

Low b-value DWI is promising for differentiation of BD and HIE with maintained perfusion. This sequence can potentially provide important prognostic information to clinicians and families and aid in critical care decisions. The utility and low-maintenance ability of b50 DWI to demonstrate perfusion abnormalities demonstrate its promise for use in the setting of cardiorespiratory arrest.

Low b-value DWI is promising for the diagnosis of brain death and HIE after cardiorespiratory arrest. This method demonstrated perfect prediction of brain death using a cutoff ADC value of < 1.085 × 10⁻³ mm²/s in the gray matter. Low b50 ADC values in the gray matter were more predictive of brain death than in the white matter.