Optimum duration of hyperventilation during electroencephalography

Abstract

Hyperventilation (HV) is carried out for 3 minutes as a standard activation procedure in most routine electroencephalographic (EEG) procedures. The cerebral blood flow (CBF) reduction and the accompanying cerebral vasoconstriction caused by HV is believed to be the mechanism of EEG activation during HV. Some advocate for 5 minutes of HV, although the optimum duration is unknown. In this study, we measured the CBF continuously over the anterior temporal lobes using subdural probes, which use thermal diffusion flowmetry to measure CBF directly from the cerebral cortex. We sought to determine the duration of HV that produces the maximum reduction in CBF during routine HV in our epilepsy monitoring unit and prolonged the procedure for an additional 2 minutes for this study. Flowtronics^® CBF probes were placed over the anterior temporal lobes in addition to the standard subdural strip placement for localization of their seizure focus in six patients who were candidates for epilepsy surgery. CBF was measured continuously for 2 minutes before and 5 minutes during HV for each patient. Time to reach maximum reduction of CBF for each attempt (11 temporal lobes) was computed. At 3 minutes, CBF reduction ranged from 11.6% to 40.0% from the pre-HV CBF level (mean 23.9%). At 5 minutes, CBF ranged from 14.3% to 42.0% (mean 25.7%). Six of the 11 measurements were steady or decreased slightly, and in the five other measurements, CBF showed a reverse trend after 3 minutes. A significant CBF reduction was attained in 3 minutes of HV in all trials. Continued HV after 3 minutes resulted in only a marginal (mean 1.8%) additional CBF reduction after 3 minutes. Thus, we propose that 3 minutes of HV is sufficient for EEG activation by the CBF criterion.

Activation with regards to electroencephalography (EEG) is defined as a group of special procedures aimed at enhancing or eliciting normal or abnormal EEG activity in the clinical electroencephalogram.¹ Many activation procedures have become part of routine EEG tests. The most used techniques are hyperventilation (HV), photic stimulation, and sleep deprivation. The precipitation of seizures by HV was well known prior to the discovery of the human EEG.^2,3 Therefore, since its introduction, EEG activation by HV has been widely used in clinical EEG laboratories. Routine in-lab activation procedures should be performed toward the end of a recording when the induced changes can be contrasted with the baseline activities.⁴ The standard HV test consists of 3 minutes of vigorous overbreathing. Some investigators have recommended more prolonged HV for 5 minutes to ensure that the physiologic changes required are induced by the procedure. Because changes in blood gases may reach their maximum changes some minutes after HV has ceased, Achenbach-Ng et al recommended continuing EEG recording for 5 minutes.⁵ The present study was designed to 1) evaluate the magnitude of cerebral blood flow (CBF) reduction during HV employing the novel technology of continuous cerebral blood flow (CCBF) measurements using thermal diffusion probes, and 2) determine the time it takes to reach trough flow by measuring CCBF directly from the normal and abnormal lateral temporal cortices of candidates for temporal lobectomy.

METHODS

Bitemporal burr holes were placed under general anesthesia for insertion of subdural strips in patients requiring invasive monitoring by subdural strip electrocorticography for localization of their epileptogenic temporal lobe. Subdural strips with four contacts were inserted in the inferomedial and posterior temporal directions, and strips with eight electrodes were inserted in the lateral-frontal direction bilaterally through temporal burr holes for localization of the seizure focus (Figure 1a). Informed consent was obtained from each study subject prior to surgery, and thermal diffusion flowmetry CBF probes for CCBF monitoring (Figure 1b) were placed in the anterior temporal directions bilaterally, when possible, in accordance with the protocol approved by the Human Subjects Committee of the University of Arizona. The use of thermal diffusion flowmetry strips allows for CCBF measurement, recording, and display on an LCD screen.⁶ A modification of the Flowtronics^® probe as shown in Figure 1c allowed for the EEG to be derived from the gold and platinum contacts used for the thermistor/thermocouple, which served as the sensor for the measurement of CBF by the Flowtronics system. This allowed for correlation of CBF with the subdural strip electrocorticogram derived from the same probes, as well as the patient’s behavioral states, not only in the interictal period, but also in the ictal, immediate, and late postictal states. CBF of these patients was monitored continuously during the hospitalization until they were taken to surgery for removal of the probes prior to their partial temporal lobectomies where indicated. While still in the epilepsy monitoring unit, the patients were asked to perform HV for an extended period of 5 minutes as tolerated as part of their daily baseline EEG.

Figure 1.

(a) Placement of the medial, lateral, and frontal subdural strips and the continuous cerebral blood flow (CCBF) probe over the lateral surface of the anterior temporal lobe on the left side, as labeled for localization of the seizure focus in our Level 4 epilepsy monitoring unit. (b) The Flowtronics^® CCBF probe showing the distal platinum and proximal gold contacts, which serve as the thermistor as well as the thermocouple (sensor) used for the measurement of CCBF as it is being inserted into the temporal burr hole. (c) The modified Flowtronics^® CCBF probe showing the addition of the EEG leads to derive EEG information from the same contacts used as the thermocouple for measuring CCBF. (This modification was courtesy to the author by Flowtronics Corporation.)

Six patients participated in the study. One was under 18 and five adults were <65 years of age. CCBF probes were available bilaterally in three patients, and three had unilateral probes. One patient with bilateral probes was studied twice. This yielded a total of 11 sets of CCBF data for analysis. The blood flow data were collected for 2 minutes before HV and for 5 minutes of HV for analysis in all six patients.

RESULTS

As shown in Figure 2, HV caused a reduction in CBF within the first minute in all but one patient. All patients showed a reduction in CBF by the end of 2 minutes. At the end of 3 minutes, the CBF reduction in all patients in the study ranged from 11.6% to 40%, with a mean reduction of 23.9% from baseline. By 4 minutes, the CBF began to rise in five temporal lobes, remained steady in five, and slightly dropped in one. At 5 minutes, all values remained below baseline, ranging from 14.3% to 42%, with a mean of 25.7%, which was only a marginal reduction by 1.8% from the value at 3 minutes. The study demonstrated that change in CBF occurred after the first minute in all patients and that maximal changes occurred by 3 minutes. Any additional change in CBF at 5 minutes was negligible. Considering the side effects that may result from HV, albeit benign in most cases, HV does not need to be prolonged beyond 3 minutes, as that time is sufficient for routine EEG activation.

Figure 2.

(a) Continuous cerebral blood flow (CBF) during 2 minutes of baseline (pre-hyperventilation [HV]) and 5 minutes of HV showing a relatively rapid reduction in CBF as HV starts. A trough level is reached by 3 minutes, with minimal changes noted in the last 2 minutes of HV. (b) Percentage reduction of CBF from baseline as HV starts with a rise (return toward) baseline in four instances after 4 minutes.

DISCUSSION

The American EEG Society recommends HV and photic stimulation as part of a standard EEG.⁴ HV has been known to produce clinical seizure activity for many years. It is often used in the clinical setting to precipitate absence seizures for diagnostic purposes, especially in pediatric practices, and is one of the most common activation procedures performed in the EEG laboratory.⁷ HV precipitates epileptiform activity and seizures in about 7% of the trials and in approximately 4% of patients with complex partial seizures.⁸ HV reduces the partial pressure of carbon dioxide by respiratory washout. HV results in respiratory alkalosis and cerebral vasoconstriction, which result in reduced CBF, lower delivery of energy substrates to the brain, and increased neuronal excitability.⁹ The vasoconstriction during HV has also been postulated to result from direct effects of hypocapnia on mesencephalic reticular formation. Routine HV is performed as a standard procedure for 3 minutes in the EEG laboratory to enhance focal slowing and other EEG changes. Some investigators have recommended more prolonged HV of 5 minutes.⁵ This study attempted to resolve this controversy. Focal slowing during HV is considered abnormal and is suggestive of underlying cortical pathology; it occurs more frequently than abnormal spike activity or seizures. Prolonged focal slowing after cessation of HV also is considered abnormal.¹⁰

The most common EEG response to HV by a normal individual is diffuse slowing.^{4,7–9,11–15} In children, the response usually begins in the occipitotemporal regions and spreads to the rest of the brain.¹⁶ With increasing age, the EEG response appears most prominently at the vertex.⁹ Yamaguchi et al demonstrated that normally, advancing age decreases the cerebral vasoconstrictive response during HV.¹⁷ On visual analysis of the EEG, the slowing disappears within 5 to 60 seconds after cessation of overbreathing. Side effects that may occur during and following HV include dizziness, numbness, headache, feeling hot or cold, blurred or double vision, a feeling of pins and needles, cramps with inability to open hands, chest tightness, lightheadedness, muscle twitching, tingling sensation, inability to think clearly, tightness around the head or neck, feelings of detachment, fatigue, apprehension, breathing difficulty, and muscular incoordination.¹¹ On rare occasions, it may lead to a syndrome of persistent HV, which promptly responds to breathing into a bag.

Aside from these relatively benign responses, HV may either accentuate an existing abnormality or unmask a latent abnormality.⁹ HV may induce focal slowing, sharp waves, or spike and slow wave discharges of epileptogenic significance. It is particularly effective in eliciting bilaterally synchronous spike wave discharges in patients with generalized epilepsies. HV is often used in the clinical setting to precipitate absence seizures for diagnostic purposes.^7–9,13 In some patients, 3 Hz spike-and-slow wave activity is detected only during HV.⁹ Dalby et al in 1969 found that patients with petit mal (absence) seizures were more sensitive to HV than non–petit mal seizures, with the incidence of spike wave paroxysms being 50% in petit mal and 25% in non–petit mal cases.¹⁸ Focal slow wave activity associated with localized lesions may be enhanced, with abnormalities in the temporal regions being more prone to enhancement than lesions elsewhere.¹⁹ Miley et al in 1977 reported that HV induced abnormal discharges in 11% of patients with complex partial seizures, of which 4.4% had clinical seizures during HV.⁸

Many have claimed that the slowing of EEG during HV is caused by ischemia secondary to the hypocapnia induced by vasoconstriction rather than by direct neuronal effects of carbon dioxide or the associated alkalosis. This interpretation was supported by subsequent demonstration of regional gray-white matter differences in functional magnetic resonance imaging signal changes during voluntary HV in seven male subjects regulated by capnometry. Researchers made measurements on multisection echo-planar magnetic resonance images and suggested that it may reflect differences in metabolic activity, vascular regulation, and/or capillary density.²⁰ However, in addition to possible hemodynamic changes, there are additional possibilities to be considered. HV is known to produce an increase in neuronal excitability. Work done on in vitro preparations suggests that neuronal hyperexcitability associated with HV is largely attributable to respiratory alkalosis, which takes place within nervous tissue during the net loss of carbon dioxide.²¹ Indeed, a study of the pH dependence of a large number of ligand- and voltage-gated channels indicated that a rise in pH will increase both intrinsic neuronal excitability and excitatory (glutamatergic) synaptic transmission while suppressing GABA A-receptor-mediated postsynaptic inhibition.^22–24 Such an increase in the general level of neuronal excitation and spontaneous firing of cortical neurons could reduce stimulus-locked synaptic events and, consequently, the maximum amplitudes of neuronal mass responses. There appears to be a variability of reactivity to cerebral vasculature in the region of the seizure focus, with responses to HV being significantly different for epileptogenic regions compared to nonepileptogenic cortex. Patel and Maulsby proposed that the vasoconstriction observed during HV is a central neurogenic response to hypocapnia at a brainstem level.⁹

In addition to hypocapnia, the response to HV is known to be affected by metabolic factors. Hypoglycemia tends to enhance the appearance of slow waves. This response may be suppressed by the administration of glucose. Other factors having relevance to this response include those of cerebral, cerebrovascular, and possibly genetic origin. EEG activation should not be carried out in patients with subarachnoid hemorrhage, ischemic stroke, intracranial hemorrhage, or significant cardiopulmonary disease. This is because HV results in significant heart rate and blood pressure increases—changes that are influenced by age, gender, smoking, and hypertension.

In conclusion, the study provided proof of reduced blood flow over the cerebral cortex by direct continuous measurement of CBF during HV. This has not been done before, to the best of our knowledge, using the technology described. Since trough flow was reached in 3 minutes and subsequent changes were minimal, one may conclude that it is sufficient to perform HV for 3 minutes in most patients for routine EEG to demonstrate CBF-related changes and hence activation. Another interesting observation that we made which also has not been documented in the literature by direct CBF measurement is the differences in flow reduction between the abnormal and normal cortices. The abnormal cortex seems to be more susceptible to the effects of hypocapnia and/or respiratory alkalosis, by responding with a greater reduction in CBF. The reduction of CBF over the abnormal cortex was twice that of the normal cortex in this small group of patients.

Disclosure statement/Funding

The authors report no funding or conflicts of interest.