Cerebrovascular autoregulation in diabetic ketoacidosis: Time to go with the (microvascular cerebral blood) flow!

In this issue of Pediatric Critical Care Medicine, Ma and coworkers report the largest prospective observational cohort study to date of cerebral blood flow autoregulation in 32 children with diabetic ketoacidosis (DKA)^X. Using transcranial doppler (TCD) ultrasound and a tilt table test to modulate blood pressure, the authors found evidence of impaired cerebral autoregulation in 40% of children with DKA assessed during the first 12-24 h, which resolved in most cases by 36-72 h after beginning insulin therapy. The incidence of impaired autoregulation was not different between those with (n = 7) and without (n = 25) evidence of cerebral edema. The authors conclude that impaired cerebral autoregulation is common early in the course of severe DKA in children.

From a technical standpoint the study was well planned and executed with early and late assessments of autoregulation changes. A single investigator performed TCD measurements, and a relatively high proportion of children (7/32) had clinical or radiological evidence of cerebral edema, although inclusion criteria for edema (CT findings that may be incidental and mental status changes explained by metabolic derangements rather than intracranial hypertension) might have resulted in falsely high estimates of clinically significant edema. Another caveat to keep in mind is that tilt testing can significantly change intracranial pressure (ICP), especially in patients with existing intracranial hypertension, making changes in cerebral perfusion pressure (CPP) uncontrollable and resulting in false autoregulation phenomena. Other factors that may limit the clinical impact of the study include uncertainty about the relationship between CBF velocity and microvascular CBF in DKA patients (discussed below), the finding that ARI < 0.4 did not discriminate children who developed cerebral edema. Moreover, TCD measurements were begun after the time that cerebral edema usually presents clinically in DKA patients. These caveats notwithstanding, the study by Ma et al.^X addresses critical and unresolved questions in the DKA literature: What is the magnitude and temporal course of the cerebral hemodynamic response to childhood DKA? Do derangements in CBF contribute to cerebral edema or ischemic brain injury in these patients?

Magnetic resonance imaging (MRI) studies in children with severe DKA show that with treatment, mean transit time of gadolinium in the brain decreases, cerebral blood volume remains stable, and ADC increases, suggesting an increase in cerebral blood flow with time as well as the presence of vasogenic edema^{1, 2}. Cerebral oximetry studies also suggest that cerebral hyperemia occurs^{3, 4}, and concomitant hypertension⁵ and increased blood-brain barrier permeability⁶ might also contribute to vasogenic edema in these patients. If so, increased cerebral blood flow might be a therapeutic target to reduce the risk of cerebral edema and death in children with severe DKA. However, other studies suggest that cerebral edema in DKA may be related to dehydration and hypocarbia, suggesting the possibility of cerebral hypoperfusion and ischemia-reperfusion brain injury⁷. In the comatose, intubated patient with DKA and intracranial hypertension, management of PaCO₂, acid-base status, blood pressure, and other modulators of CBF may critically impact neurological outcome⁸. Thus, understanding CBF dynamics is an important research goal for this patient population.

To place the study by Ma et al. in context it is important to recognize that studies using direct quantitative methods to assess microvascular CBF in DKA patients are completely lacking. Gold standard techniques for assessment of CBF after acute hemorrhagic or traumatic brain injury, such as xenon CT^{9, 10}, SPECT, and positron emission tomography¹¹, have not been reported in children or adults with DKA. Lack of basic knowledge about the temporal course and magnitude of microvascular cerebral blood flow casts significant uncertainty on interpretation of CBF autoregulation studies using TCD in DKA patients. A basic assumption of TCD studies is that changes in CBF in the microcirculation are reflected in changes in blood velocity in large conductance vessels. However, studies in patients with subarachnoid hemorrhage show little or no relationship between local CBF assessed by xenon CT^{9, 10} or PET¹¹ and predicted values based on TCD findings. In SAH patients with high velocity TCD signals predictive of vasospasm, some had normal or even hyperemic cerebral blood flow^{9, 11}. Moreover, data are lacking regarding the possibility of changes in diameter of large conductance vessels during the course of DKA. The presence of vasospasm or vasoplegia would certainly impact interpretation of TCD data. In the study by Ma et al., mean arterial blood pressure remained unchanged and PvCO₂ increased significantly over time, yet MCA blood velocity remained unchanged between the two measurement periods. Similar results were obtained by Roberts et al.¹² who found no change in cerebral blood flow velocity in the MCA despite increasing PaCO₂ with treatment of DKA. These findings were interpreted as normal to increased CBF, but in fact cannot be fully understood without direct quantitation of microvascular CBF. This is more than a theoretical problem because cerebrovascular reactivity index measurements, incorporating ICP waveforms as well as assessment of CBF, may determine optimal cerebral perfusion pressure and improve outcome in patients with severe traumatic brain injury¹³, and the same might be true for patients with DKA and intracranial hypertension as well.

Diffuse correlation spectroscopy (DCS) is one technology that shows promise of bedside assessment of microcirculatory cerebral blood flow. DCS measures the temporal intensity fluctuations of photons caused by moving red blood cells as light travels from a source emitter, through the brain, and back to photodetectors. The temporal intensity autocorrelation function of the detected light is computed, and correlation diffusion theory is used to obtain a tissue blood flow index (cm²/s). Previous studies in brain injured adults show good correlation between relative CBF changes obtained with DCS and xenon CT¹⁴, and microcirculatory CBF assessed by DCS correlates well with MCA blood velocity assessed by TCD in healthy volunteers¹⁵, suggesting that DCS alone or combined with TCD might be used to better understand cerebrovascular autoregulation in DKA. If so, such technology might bring us one step closer to noninvasive real time monitoring of microvascular CBF reactivity in children with DKA, as well as other CNS diseases and injuries for which CBF measurements outside the PICU may not be safe or feasible.