Investigation of Whole-Brain White Matter Identifies Altered Water Mobility in the Pathogenesis of High-Altitude Headache

Abstract

Elevated brain water is a common finding in individuals with severe forms of altitude illness. However, the location, nature, and a causative link between brain edema and symptoms of acute mountain sickness such as headache remains unknown. We examined indices of brain white matter water mobility in 13 participants after 2 and 10 hours in normoxia (21% O₂) and hypoxia (12% O₂) using magnetic resonance imaging. Using a whole-brain analysis (tract-based spatial statistics (TBSS)), mean diffusivity was reduced in the left posterior hemisphere after 2 hours and globally reduced throughout cerebral white matter by 10 hours in hypoxia. However, no changes in T₂ relaxation time (T₂) or fractional anisotropy were observed. The TBSS identified an association between changes in mean diffusivity, fractional anisotropy, and T₂ both supra and subtentorially after 2 and 10 hours, with headache score after 10 hours in hypoxia. Region of interest-based analyses generally confirmed these results. These data indicate that acute periods of hypoxemia cause a shift of water into the intracellular space within the cerebral white matter, whereas no evidence of brain edema (a volumetric enlargement) is identifiable. Furthermore, these changes in brain water mobility are related to the intensity of high-altitude headache.

Keywords

altitude illness diffusion tensor imaging hypoxia magnetic resonance imaging

INTRODUCTION

Prolonged hypoxia causes a number of pathologic symptoms including headache, nausea, fatigue, gastrointestinal discomfort, and sleep disturbances that are collectively known as acute mountain sickness.¹ In fact, some individuals have more serious neurologic symptoms including hallucinations, ataxia, and unconsciousness that in some cases can lead to death (high-altitude cerebral edema). ² Both syndromes have been associated with alterations in brain water.^2–4 However, although focal (splenium corpus callosum) vasogenic (extracellular) cerebral edema is commonly found in patients having high-altitude cerebral edema,² the location, nature, and causative link between brain edema and symptoms of acute mountain sickness are currently a subject of intense debate.⁵

Early work examining the impact of hypoxia on brain edema and acute mountain sickness reported no change in brain water content on exposure to high altitude.^6,7 However, recent magnetic resonance imaging studies using larger sample sizes or more sensitive techniques (using higher magnetic fields) have identified the presence of vasogenic (extracellular) edema and intracellular swelling after 6 hours⁴ and 16 hours³ of exposure to simulated high altitude (4,500 m). More specifically, these investigators report an inverse correlation between apparent diffusion coefficient (ADC, a marker of water mobility whereby a decrease would indicate intracellular swelling) and acute mountain sickness, which was interpreted as evidence for the involvement of intracellular swelling in the development of acute mountain sickness.

Despite advancing our understanding of cerebral edema during periods of hypoxemia and acute mountain sickness, critical analysis of both these studies is warranted. First, although Schoonman et al⁴ report a correlation between intracellular swelling and acute mountain sickness, only 3 of 9 participants showed a decreased ADC (−4%), whereas 3 of 6 participants with acute mountain sickness actually showed an increase in ADC (+ 4%). Second, Kallenberg et al³ report differences in the change in ADC between individuals with and without acute mountain sickness, but individuals supposedly free of acute mountain sickness did report meaningful symptoms and in both groups acute mountain sickness symptoms were beginning to resolve when the measurements were obtained. Thus, interpreting the relationship between brain water and acute mountain sickness in both these investigations is difficult. This being said, a recent investigation⁸ using similar techniques to examine brain water after 2 and 7 days at a terrestrial attitude of 3,800 m has largely supported these early findings.

In addition, all previous investigations examined indices of brain edema in established acute mountain sickness, which makes a causative link between brain edema and illness difficult to determine. Also, on a methodological note, previous studies have predominantly focused on manually drawn prespecified and prelocalized regions or features of interest to determine the location of brain edema and used the ADC in combination with T₂-weighted images to indicate intracellular swelling. Manually drawn regions of interest are favorable for a hypothesis-driven analysis and increased statistical power but they are subject to human error and do not investigate the whole brain. Furthermore, the ADC represents a single average diffusion coefficient, which is overly simplistic when examining anisotropic diffusion as occurs within brain white matter.⁹ A more detailed approach to examine brain tissue microstructure throughout the whole brain is therefore warranted.

Recent advances in neuroimaging techniques such as diffusion tensor imaging allow the investigation of brain water mobility within three-dimensional space. This concept is important, as water mobility within tissues is not necessarily the same in all directions. Diffusion tensor imaging data can be analyzed in a number of ways to reveal intricacies in brain water mobility. For example, the diffusion parameters calculated by tensor techniques can be used to describe ellipsoids, with the mean diffusivity characterizing the average ellipsoid size at each imaging voxel, which is independent of tissue directionality but dependent on cellular size and integrity and the location of water molecules being measured (analogous to the ADC), whereas fractional anisotropy describes the degree of diffusion anisotropy within space (the higher the anisotropy, the more eccentric the ellipsoid) and thus reflects tissue structural arrangement and integrity.

Movement of water from the extracellular space to the intracellular space would be expected to decrease mean diffusivity and increase fractional anisotropy, because of the more restricted and anisotropic diffusion environment found intracellularly. Vice versa, movement of water out of the cells would lead to a decrease in fractional anisotropy and an increase in mean diffusivity. However, only together with an increase in T₂ relaxation time (T₂) would changes in these indices indicate an increase in brain water (edema) rather than water redistribution within the brain. Importantly, each diffusion tensor imaging index can be analyzed using a region of interest analysis and whole-brain approach via tract-based spatial statistics (TBSS). Therefore, the characterization and severity of altered brain water mobility can be examined with exposure to hypoxia and in individuals who report symptoms of acute mountain sickness.

As such, this study aimed to examine brain water mobility and content after exposure to 2 and 10 hours of hypoxia (12% O₂). Subsequently, we explored the relationship between the development of a specific acute mountain sickness symptom, headache, with changes in brain water mobility and content after 2 hours (presymptomatic) and 10 hours (symptomatic) in hypoxia. Given the lack of molecular evidence for breakdown of the blood–brain barrier in hypoxia,^10,11 but evidence for the accumulation of both extracellular and intracellular water,^3,4,8 we hypothesized that early hypoxia (2 hours) will be characterized by a reduction in mean diffusivity and increased fractional anisotropy but no change in T₂ in line with a fluid shift into the intracellular space. In contrast, more prolonged hypoxia (10 hours) will be associated with reduced mean diffusivity but increased fractional anisotropy and T₂, which would suggest an increase in brain edema. Finally, we hypothesized that the increase in intracellular swelling after 2 hours and the osmotically obligated development of brain edema after 10 hours will be related to headache intensity after 10 hours by its expected effect on increasing brain volume and likely intracranial pressure.

MATERIALS AND METHODS

Participants

A total of 13 males (age = 26 (s.d. 6) years; body mass = 77 (s.d. 12) kg; height = 182 (s.d. 9) cm) volunteered to take part in this study. Informed consent was obtained after each participant was given an explanation of the experimental protocol and fully understood the possible risks involved in taking part in the study. Exclusion criteria were any clinically diagnosed primary headache disorder and ascent to altitude above 1,500 m in the previous 6 months. This study was approved by the Institutional Ethics Committee and conformed to the Declaration of Helsinki (2008).

Experimental Design

Participants reported to the laboratory at 0700 hours on two separate occasions separated by at least 5 days. Participants were asked to refrain from alcohol consumption and exhaustive exercise for a period of 48 hours before each visit. Furthermore, in the 24 hours before and during each testing session, participants were instructed to drink 40 mL/kg of water to maintain an adequate hydration, and were instructed to consume an ad libitum diet, which was recorded and repeated in both trials. Controlled fluid and nutritional intake was important to minimize the possibility of systemic dehydration affecting central fluid dynamics.

Each trial consisted of a 10-hour exposure to normobaric normoxia (21% O₂) or normobaric hypoxia (12% O₂) in a temperature-(23°C) and humidity (40%)-controlled environmental chamber. Participants were assigned to each exposure in a randomized order (http://www.randomization.com) and were masked to the experimental conditions (hypoxia or normoxia). Participants underwent magnetic resonance imaging after 2 and 10 hours in both trials. Importantly, participants maintained breathing the appropriate experimental gas (21% O₂ or 12% O₂) during transportation to and throughout the magnetic resonance imaging. All magnetic resonance imaging sequences were obtained after a 5-minute period of supine rest.

Cardiorespiratory Variables

Oxygen saturation and heart rate were monitored continuously in the environmental chamber (TM-2564GP; A&D Medical, San Jose, CA, USA) and brachial systolic and diastolic blood pressure measures were obtained in duplicate by an automated inflating cuff after 2 and 10 hours in the environmental chamber (TM-2564GP; A&D Medical). End-tidal carbon dioxide was sampled from the facemask during each magnetic resonance imaging session and analyzed by fast responding gas analyzer (I.R. gas analyzer, PA404; Servomex, Sussex, UK). Cardiovascular variables were compared using a 2 (time, 2 versus 10 hours) × 2 (altitude, normoxia versus hypoxia) repeated measures analysis of variance. Follow-up comparisons were conducted using Tukey's test.

Altitude Illness

Acute mountain sickness was recorded using the Lake Louise Questionnaire¹² and headache intensity was scored on a horizontal 0-100 mm line (visual analogue scale) with verbal anchors at opposing ends (none and severe^13,14) after 2 and 10 hours in both trials. Participants were asked to place a mark on the 100 mm line at the point that they perceived reflected their current headache intensity. Acute mountain sickness and high-altitude headache were compared using a 2 (time, 2 versus 10 hours) × 2 (altitude, normoxia versus hypoxia) repeated measures analysis of variance. Follow-up comparisons were conducted using Tukey's test.

Magnetic Resonance Imaging Acquisition and Postprocessing

All magnetic resonance imaging sequences were conducted on a 3 tesla magnetic resonance imaging scanner (Phillips Achieva, Philips Healthcare, Eindhoven, The Netherlands) using a 16-channel head and neck coil. All imaging sequences were acquired with sensitivity encoding for fast magnetic resonance imaging (SENSE).

Acquisitions

Brain water content. As a measure of brain water content, the monoexponential T₂ was calculated for the whole brain via a multiecho sequence (TE, 20, 40, 60, 80, and 100 ms) within the axial orientation. Imaging parameters include field of view 240 × 240 mm²; slice thickness 2.0 mm; acquisition matrix 240 × 240; voxel dimension 1.0 × 1.0 × 2.0 mm; TR 7,297, flip angle 90°, and an acquisition time of 5 minutes 40 seconds.

Brain water mobility. Cerebral water mobility was calculated using a diffusion-weighted, spin-echo, single-shot, echo-planar imaging sequence. Imaging parameters include 34 gradient directions with b-values set at 0 and 800 s/mm², field of view 220 mm²; slice thickness 1.5 mm; number of slices 64; acquisition matrix 112 × 110; voxel dimension 1.5 × 1.5 × 1.5 mm; TE 70; TR 8,552, flip angle 90°, and an acquisition time of 15 minutes 6 seconds.

Postprocessing

Initially, diffusion tensor imaging images were processed using TBSS, part of FSL 4.1.9 (FMRIB, Oxford, UK).¹⁵ Using the Diffusion Toolbox (FDT v2.0, FMRIB), each diffusion-weighted volume was affine-aligned to its corresponding b0 image to correct possible motion artifacts and eddy current distortion. Subsequently, each eigenvector and eigenvalue (λ1, λ2, and λ3) was generated using the DTIfit algorithm (FMRIB). The separate eigenvalues were used to calculate mean diffusivity (MD; Equation (1)) and fractional anisotropy (FA; Equation (2)).

M D = \frac{λ_{1} + λ_{2} + λ_{3}}{3}

(1)

F A = \frac{1}{\sqrt{2}} \sqrt{\frac{{(λ_{1} - λ_{2})}_{2} + {(λ_{2} - λ_{3})}_{2} + {(λ_{1} - λ_{3})}_{2}}{λ_{1}^{2} + λ_{2}^{2} + λ_{3}^{2}}}

(2)

To create a fractional anisotropy skeleton for both region of interest and voxel-wise analysis, a study-specific target image was initially identified by nonlinearly registering all participants and time points to one another. Subsequently, this representative image was affine-aligned to the MNI152 standard space and every image transformed to the 1 × 1x 1 mm target MNI152 space. The mean fractional anisotropy skeleton was then thresholded to fractional anisotropy of ≥ 0.2 to include major white matter tracts but exclude peripheral and intersecting tracts and the possibility of partial volume effects. Finally, the mean diffusivity and T₂ calculated images were nonlinearly registered to the original fractional anisotropy skeleton. It is noteworthy that one participant was removed from the investigation after the first 2 hours of magnetic resonance imaging scanning session because of severe altitude illness and one participant's T₂ image was considered to contain too much movement artifact for voxel-wise analysis. Thus, a total of 12 participants were retained for analysis of mean diffusivity and fractional anisotropy and 11 participants for statistical analysis of T₂ using TBSS.

Tract-Based Spatial Statistics Analysis: Whole Brain

To examine the effect of hypoxia on brain water content and mobility across the whole brain, cross-subject voxel-wise statistical analysis of T₂, mean diffusivity, and fractional anisotropy were carried out using TBSS,¹⁵ part of FSL 4.1.9.¹⁶ Permutation-based nonparametric paired t-tests (5,000 permutations) were calculated to compare each hypoxic time point with its corresponding normoxic counterpart (e.g., normoxia 2 hours versus hypoxia 2 hours) for each diffusion index (mean diffusivity and fractional anisotropy) and the corresponding T₂ image using Randomize v2.1, part of FSL 4.1.9 (FMRIB). Threshold-free cluster enhancement (P<0.05 corrected for multiple comparisons across space) was used to define clusters without the need to arbitrary select cluster size threshold. Furthermore, to explore the relationship between brain water mobility and high-altitude headache, skeletonized T₂, mean diffusivity, and fractional anisotropy images (after 2 and 10 hours) were correlated, using voxel-wise analysis, with headache score after 10 hours at high altitude using Randomize v2.1. Again threshold-free cluster enhancement was used to define significant clusters. Importantly, the 2-hour time point represents a presymptomatic period, and thus correlating with headache intensity after 10 hours aids in interpretation of cause and effect. Relationships between arterial oxygen saturation and end-tidal carbon dioxide with brain water mobility were also assessed in a similar manner.

Region of Interest-Based Analysis: Mean of All Regions of Interest

Although TBSS examines the whole brain, a limitation is its conservative statistical approach because of the correction for family-wise error across all voxels. Thus, after applying the TBSS preprocessing steps above, the complementary traditional approach of drawing predefined regions of interests was undertaken. Predefined skeleton-based regions of interest were drawn for mean diffusivity and fractional anisotropy at each time point within the white matter of the frontal lobe, posterior white matter, centrum semiovale, genu and splenium of the corpus callosum, anterior limb of the internal capsule, and posterior cerebellar peduncle (Analyze-Direct, Overland Park, KS, USA; see Figure 1). The T₂ images were transferred onto an offline workstation whereby the regions of interests were manually drawn for each time point. An investigator masked to the experimental conditions and altitude illness score drew all regions of interest. To assess overall change in brain white matter water mobility, the mean of all regions of interest (mROIs) were compared using a 2 (time, 2 versus 10 hours) × 2 (altitude, normoxia versus hypoxia) repeated measures analysis of variance. Follow-up comparisons were conducted using Tukey's test. The relationship between overall brain water mobility and high-altitude headache was then assessed by correlating the change in T₂, mean diffusivity, and fractional anisotropy (using mROI) after 2 and 10 hours with high-altitude headache intensity after 10 hours by Pearson's r.

Figure 1.

Representative masks for the skeletonized-based regions of interest overlaid on mean fractional anisotropy skeleton (green). Masks include frontal lobe white matter (red), centrum semiovale (blue, plate A), posterior white matter (purple), genu (yellow) and splenium of the corpus callosum (orange, plate B), anterior limb of the internal capsule (brown, plate C), and posterior cerebellar peduncle (magenta, plate D).

Region of Interest-Based Analysis: Specific Regions

To further aid in the interpretation of the above analyses, exploratory paired t-tests were conducted for each brain region between each time point (i.e, normoxic 2 hours versus hypoxia 2 hours) to explore which brain regions contributed to the change in overall brain water mobility. Although this method is subject to type I error, it was thought essential to include as a secondary analysis to highlight the possible regional effects of hypoxia.

Values are means and s.d. and all statistical procedures were carried out on SPSS version 19 for Macintosh (Statistical Package for Social Sciences, IBM, Chicago, IL, USA). Statistical significance was accepted when P ≤0.05 and trends were stated when P ≤0.10.

RESULTS

Cardiorespiratory and Illness Responses to Altitude

As expected, exposure to hypoxia decreased oxygen saturation (main effect of altitude, F = 644.16, P< 0.001) and end-tidal carbon dioxide (main effect of altitude, F = 5.08, P = 0.05) and increased heart rate (main effect of altitude, F = 34.63, P < 0.001) with a trend toward an increase in mean arterial blood pressure (main effect of altitude, F = 3.69, P = 0.08, see Table 1). Acute mountain sickness (altitude by time interaction, F = 7.41, P = 0.02), and high-altitude headache (altitude by time interaction, F = 13.33, P<0.001) were significantly elevated after 10 hours in hypoxia.

Table 1

Physiologic variables and altitude illness in normoxia and hypoxia

	Normoxia		Hypoxia (12% O₂)

	2 Hours	10 Hours	2 Hours	10 Hours
Oxygen saturation (%)	99 (1)	99 (1)	81 (4)^*	81 (1)^*
End-tidal carbon dioxide (%)	4.7 (0.7)	4.7 (0.3)	4.5 (0.9)	3.9 (0.6)
Heart rate (beats/min)	59 (14)	62 (15)	74 (18)^*	78 (20)^*
Mean arterial pressure (mm Hg)	81 (9)	83 (7)	88 (14)	87 (8)
Lake Louise questionnaire (a.u.)	0 (0)	1 (1)	1 (1)	4 (3)^*
High-altitude headache (a.u.)	2 (5)	6 (13)	5 (7)	41 (24)^*

a.u., arbitrary units; O₂, oxygen.

^*P≤0.05 compared with the respective sea level control, by post hoc Tukey's test. Note the decrease in end-tidal carbon dioxide did not reach statistical significance using follow-up Tukey's test. Values are means (s.d.).

Tract-Based Spatial Statistics Analysis: Whole Brain

When examining the impact of hypoxia on brain water mobility across the whole brain, voxel-wise analysis using TBSS revealed a significant reduction in mean diffusivity throughout the left posterior hemisphere after 2 hours in hypoxia. Furthermore, after 10 hours in hypoxia, TBSS identified extensive reductions in mean diffusivity supratentorially throughout the cerebral white matter (see Figure 2). In contrast, no significant differences in T₂ or fractional anisotropy were observed at either time point using this technique.

Figure 2.

Voxel-wise, whole-brain analysis showing decreased mean diffusivity after 2 hours (upper panel) and 10 hours (lower panel) of exposure to 12% oxygen (O₂). Significance maps (red-yellow, corrected P≤0.05) are overlaid on the study-specific mean fractional anisotropy tract-based spatial statistics (TBSS) skeleton (green) and T_q-weighted MNI152 standard image for identification of affected anatomy. The TBSS statistics have been thickened to aid interpretation of results.

Region of Interest-Based Analysis: Mean of All Regions of Interest

As shown in the upper panel of Table 2, using the mROI, hypoxia caused a global reduction in mean diffusivity after only 10 hours (altitude by time interaction, F = 38.86, P<0.01). Meanwhile, fractional anisotropy for the mROI was slightly elevated after both 2 and 10 hours (main effect of altitude, F = 25.00, P<0.01). Overall, exposure to hypoxia did not effect T₂ within the cerebral white matter (altitude by time interaction, F = 0.15, P = 0.7).

Table 2

Summary of T₂, mean diffusivity, and fractional anisotropy for mean of all regions of interest (mROIs) and each region of interest in normoxia and hypoxia

Region of interest	Trial	T₂ (ms)		Mean diffusivity (x 10⁻⁶cm²/s)		Fractional anisotropy

		2 Hours	10 Hours	2 Hours	10 Hours	2 Hours	10 Hours
Primary whole-brain region of interest-based analysis
mROI	Normoxia	78 (3)	78 (3)	598 (2)	634 (2)	0.57 (0.02)	0.56 (0.01)
	Hypoxia	78 (3)	78 (3)	589 (3)	621 (2)^*	0.58 (0.02)^*	0.57 (0.01)^*
Significance		Effect of hypoxia F = 0.28, P = 0.60		Hypoxia x time interaction, F= 38.86, P<0.01		Effect of hypoxia, F = 25.00, P<0.01
Secondary region of interest-based analysis
Centrum semiovale	Normoxia	75 (4)	78 (4)	661 (3)	658 (3)	0.34 (0.03)	0.36 (0.04)
	Hypoxia	75 (4)	77 (3)	641 (3)^a	642 (4)^a	0.35 (0.04)	0.38 (0.04)^a
Frontal lobe	Normoxia	80 (4)	78 (4)	561 (3)	636 (3)	0.51 (0.05)	0.53 (0.06)
	Hypoxia	77 (4)^a	77 (4)	559 (3)	635 (4)	0.51 (0.06)	0.53 (0.04)
Posterior white matter	Normoxia	84 (3)	89 (3)	477 (3)	738 (4)	0.60 (0.02)	0.57 (0.02)
	Hypoxia	84 (4)	85 (5)^a	465 (3)^b	717 (4)^a	0.62 (0.03)^a	0.57 (0.02)
Genu	Normoxia	72 (5)	72 (6)	785 (5)	737 (6)	0.69 (0.05)	0.68 (0.05)
	Hypoxia	71 (6)	70 (5)	828 (7)^b	742 (6)	0.68 (0.04)	0.67 (0.04)
Splenium	Normoxia	78 (5)	80 (8)	564 (2)	517 (2)	0.82 (0.03)	0.71 (0.03)
	Hypoxia	82 (7)	79 (5)	566 (4)	493 (2)^a	0.84 (0.04)^b	0.72 (0.05)
Internal capsule	Normoxia	67 (8)	67 (6)	534 (2)	542 (3)	0.58 (0.03)	0.60 (0.03)
	Hypoxia	68 (8)	68 (7)	521 (2)^b	537 (2)	0.59 (0.03)^b	0.61 (0.03)
Cerebellum	Normoxia	87 (6)	84 (6)	603 (5)	608 (3)	0.46 (0.03)	0.47 (0.03)
	Hypoxia	86 (5)	87 (8)	542 (11)^b	579 (6)	0.49 (0.07)^b	0.48 (0.05)

^*P≤0.05 compared with sea level control, by post hoc Tukey's test. Values are means (s.d.). PP0.05 and P<0.1 compared with sea level control using exploratory t-test.

Region of Interest-Based Analysis: Specific Regions

As shown in Table 2, lower panel, of the regions of interest measured, exploratory t-tests revealed that the global reduction in mean diffusivity after 10 hours was predominantly driven by decreases in the centrum semiovale (Δ −15.73 × 10⁻⁶cm²/s, t = 2.83, P =0.02), posterior white matter (Δ = 20.59 × 10⁻⁶cm²/s, t =2.67, P = 0.02), and splenium of the corpus callosum (Δ −23.78 × 10⁻⁶cm²/s, t = 3.63, P = 0.04). Similarly, exploratory t-test revealed the possibility that mean diffusivity was decreased after only 2 hours in hypoxia in specific brain regions including centrum semiovale (Δ20.05 × 10⁻⁶cm²/s, t = 4.55, P<0.01), posterior white matter (Δ −11.84 × 10⁻⁶cm²/s, t = 2.12, P = 0.06), internal capsule (Δ −12.85 × 10⁻⁶cm²/s, t = 2.05, P = 0.06), and cerebellum (Δ −60.66 × 10 ⁻⁶cm²/s, t = 1.81, P = 0.1). Similar to mean diffusivity, exploratory t-tests revealed that fractional anisotropy was significantly elevated in the centrum semiovale (Δ0.02, t = 2.25, P = 0.05) and posterior white matter (A0.02, t = 3.04, P = 0.01), whereas a trend was observed in splenium of the corpus callosum (A0.1, t = 1.70, P = 0.1), internal capsule (A0.01, t = 1.80, P = 0.1), and cerebellum (A0.03, t = 1.71, P = 0.1, see Table 2). Finally, exploratory paired t-tests revealed a focal reduction in T₂ within the frontal lobe (Δ −2.68 ms, t = 2.84, P = 0.01) and posterior white matter (Δ −3.30 ms, t = 2.25, P = 0.05) after 2 and 10 hours, respectively.

Relationship with High-Altitude Headache

Suggestive of intracellular involvement in altitude illness, voxel-wise statistical analysis of the whole brain revealed that T₂, mean diffusivity, and fractional anisotropy after 2 hours (presymptomatic, see Figure 3) and 10 hours (symptomatic, see Figure 4) in hypoxia was correlated with the intensity of high-altitude headache experienced after 10 hours in hypoxia. Interestingly, altered areas of white matter water mobility were detected both supra- and subtentorially using this technique. Using the mROI identified a positive relationship between the change in T₂ (r = 0.69, P = 0.01), a trend toward a positive relationship with a change in mean diffusivity (r = 0.51, P = 0.09), but no relationship with fractional anisotropy (r = 0.14, P = 0.7, see Figure 5) and headache intensity after 10 hours. No relationship was found between T₂ (r = 0.30, P = 0.4), mean diffusivity (r = 0.36, P = 0.2), or fractional anisotropy (r = 0.21, P = 0.5) after 2 hours in hypoxia using the mROI and headache intensity after 10 hours.

Figure 3.

Voxel-wise statistical analysis of the relationship between T₂ (top panel, A), mean diffusivity (middle panel, B), and fractional anisotropy (bottom panel, C) after 2 hours in hypoxia and headache score after 10 hours in hypoxia. Significance maps (red-yellow, corrected P≤0.05) are overlaid on the study-specific mean fractional anisotropy tract-based spatial statistics (TBSS) skeleton (green) and T₁-weighted MNI152 standard image for identification of affected anatomy.

Figure 4.

Voxel-wise statistical analysis of the relationship between T₂ (top panel, A), mean diffusivity (middle panel, B) and fractional anisotropy (bottom panel, C) after 10 hours in hypoxia and headache score after 10 hours in hypoxia. Significance-maps (red-yellow, corrected P≤0.05) are overlaid on the study-specific mean fractional anisotropy tract-based spatial statistics (TBSS) skeleton (green) and T₁-weighted MNI152 standard image for identification of affected anatomy.

Figure 5.

Relationship between the change in mROI of T₂ (upper panel), mean diffusivity (middle panel), and fractional anisotropy (bottom panel) after 10 hours in hypoxia with headache intensity after 10 hours in hypoxia. a.u., arbitrary units; HAH, high-altitude headache; mROI, mean of all regions of interest.

Relationships with Oxygen Saturation and End-Tidal Carbon Dioxide

Tract-Based Spatial Statistics identified that the decrease in oxygen saturation after 2 hours in hypoxia was related to the level of intracellular swelling in some brain regions (mean diffusivity and fractional anisotropy) but not T₂ after 2 hours in hypoxia (see Supplementary Figure S1). However, these relationships were not observed after 10 hours in hypoxia (see Supplementary Figure S2). Despite only a modest change in end-tidal carbon dioxide in some individuals, we did observe some brain regions where T₂, mean diffusivity, and fractional anisotropy were related to end-tidal carbon dioxide tension after only 2 hours in hypoxia (see Supplementary Figure S3). Interestingly, despite the general decrease in end-tidal carbon dioxide in most subjects after 10 hours in hypoxia, these relationships were still apparent for mean diffusivity and fractional anisotropy, although not for T₂ (see Supplementary Figure S4).

DISCUSSION

The present findings highlight for the first time that acute hypoxemia causes extensive alterations in cerebral white matter water mobility. Indeed, using a combination of whole brain and prospectively drawn region of interest analyses, these data identify that as early as 2 hours after the onset of arterial hypoxemia, intracellular white matter water mobility is reduced and is further compounded by a prolonged exposure to hypoxia.

The observed perturbations in brain water mobility are indicative of diffuse intracellular swelling and are related to the intensity of high-altitude headache experienced after 10 hours in a hypoxic environment.

Kallenberg et al³ were the first to identify the presence of regionally specific intracellular swelling (using magnetic resonance imaging) in certain individuals after 16 hours in a hypoxic (12% O₂) environment. Interestingly, their data suggest that hypoxia produces a mild intracellular swelling, localized to the Genu and Splenium of the corpus callosum, an area that seems to have a specific predilection with the development of high-altitude cerebral edema.² In contrast, a subsequent investigation using a more sensitive technique (3 tesla magnetic resonance imaging), reported an average increase in ADC in combination with an average increase in T₂-weighted signal intensity (indicative of intracellular swelling on top of extracellular edema) after 6 hours of hypoxia.⁴

Despite using a similar sample size and high magnetic field (3 tesla), we were unable to detect a global increase in T₂ using either TBSS or predefined regions of interest. Using exploratory t-tests, we actually identified a focal decrease in T₂ within the frontal lobe and posterior white matter after 2 and 10 hours, respectively, suggestive of a regional decrease in brain water within white matter or a compartmental shift of water from the extracellular space to the intracellular space, where T₂ is lower. Although contradictory to the conventional hypothesis, our findings of focally reduced brain water during acute normobaric hypoxia do agree with the recent findings of reduced T₂-weighted signal intensity during prolonged residence (7 days) at terrestrial altitude.⁸

Using a combination of TBSS and ROI analyses, we identified that 2 hours in hypoxia caused a focal reduction in mean diffusivity within the posterior left hemisphere (TBSS) that propagated throughout cerebral white matter by 10 hours (using both TBSS and skeletonized regions of interest). After both 2 and 10 hours in hypoxia, concomitant increases in fractional anisotropy were also observed using a region of interest analysis. Thus, our data are the first to identify that acute hypoxia is actually associated with profound alterations in diffusion parameters that are suggestive of a redistribution of water to within the intracellular space of glia cells. Furthermore, these data identify similar locations of altered white matter water mobility as those observed previously in individuals after prolonged sojourns to high altitude^8,17,18 and in individuals native to high altitude.¹⁹ It should be mentioned that by directly calculating T₂ and mean diffusivity, we avoided any potential confusion that ‘T₂-shine through’ can cause on raw diffusion-weighted images. Our findings of no overt increases in T₂ during hypoxia preclude the possibility of T₂ shine through on the raw diffusion-weighted image, which normally occurs when there is an increase in T₂ in the tissue.

As such, the contradictory observation in T₂ between the current investigation and that of others⁴ is likely attributable to other differences in measurement technique. For example, Schoonman et al⁴ used a T₂-weighted (B0) image that is highly susceptible to T₂^* effects. For example, changes in blood flow may lead to an apparent change in signal intensity because of the blood oxygen level-dependent effect. Conversely, we calculated the true mono-exponential T₂ decay rate, which would be more sensitive to changes in water content if they existed. Furthermore, using a separate magnetic resonance technique, we identified a decrease in total white matter volume during the early period of exposure to hypoxia (Lawley, Oliver, Mullins and Macdonald, unpublished data), which is in opposition to that expected with vasogenic edema. An alternative physiologically plausible explanation is that the current data represent the early effects of moderate hypoxemia (SaO₂ ∼81%), whereas the investigation of Schoonman et al,⁴ although shorter in duration, signifies a more advanced state because of the use of a slightly more aggressive protocol (SaO₂ ∼75% to 80%) and because participants remained in the supine position for the entire experiment, a position that is more likely to cause larger increases in intracranial pressure.^20,21

The lack of molecular evidence for breakdown of the blood–brain barrier in similar models to the current investigation^10,11 and the accumulation of intracellular water identified herein, alongside extracellular edema identified previously,⁴ hints toward the premorbid onset of regionally specific intracellular water. This may be followed by intracellular accumulation of osmotically active ions and a transcapillary flux of Na + in an attempt to rebalance extracellular ion homeostasis. Subsequently, an expansion of the extracellular space occurs (vasogenic edema), which is only observed with more prolonged or severe hypoxic exposure in conjunction with high-altitude cerebral edema. In opposition to this hypothesis, Hunt et al⁸ identified no evidence linking cerebral edema to acute mountain sickness susceptibility after 2 days at 3,800 m, although this result is not surprising given the low incidence of high-altitude cerebral edema at altitudes between 2,500 and 5,000 m.²²

Given that a central tenant of the intracellular swelling explanation is disruption of cellular membrane Na + /K + ATPase, it should be remembered that global increases in cerebral blood flow^23,24 and maintained or increased global cerebral metabolic rate of oxygen^11,25,26 have been observed. Thus, it is possible that the reduction in white matter water mobility as observed herein is the consequence of regionally reduced white matter blood flow and oxygen delivery because of transient or persistent elevations in intracranial pressure. A further alternative or complementary hypothesis is that white matter blood flow is reduced at the expense of increased gray matter perfusion, a so-called vascular steal phenomenon. Previous experiments have identified a paradoxical reduction in white matter blood flow during periods of hypercapnia in healthy individuals.²⁷ Whether such a mechanism exists to preserve gray matter blood flow and maintain neuronal activity during prolonged periods of hypoxemia and hypocapnia remains unknown. However, the observed relationships between oxygen saturation and more convincingly end-tidal carbon dioxide and indices of intracellular swelling do lend indirect support to this hypothesis. Nonetheless, future experiments directly evaluating white and gray matter perfusions over prolonged periods of hypoxia are required to substantiate this theory.

Using voxel-wise statistical analysis of the whole brain, we identified significant regional relationships between T₂, mean diffusivity, and fractional anisotropy after 2 and 10 hours in hypoxia, and headache intensity after 10 hours in hypoxia. To the best of our knowledge, this is the first experiment showing that indices of brain water mobility during the hypoxic presymptomatic period are related to the subsequent development of headache after 10 hours in hypoxia. Indeed, this time series design provides convincing evidence of a relationship between early changes in white matter water mobility and acute altitude pathology.

However, close inspection of the data suggests that the relationship between white matter water mobility and high-altitude headache is complex. For example, when exploring the relationship between a global indicator of brain water mobility (mROI) and headache intensity, we were only able to identify a significant relationship between the change in T₂ (alongside a similar trend for mean diffusivity) after 10 hours and headache score after 10 hours. This may, in part, be because of the mROI analysis using an average of all brain regions, even though some regions were unaffected by hypoxia. Furthermore, although the global increase in T₂ and preservation of mean diffusivity were predominantly associated with severe cases of high-altitude headache (suggestive of brain edema), there were a number of individuals who had decreased T₂ and mean diffusivity and yet they still reported headaches of a similar moderate intensity. This observation is strikingly similar to those obtained previously.^3,4 Despite the complexity of these acute responses, prolonged exposures have also implicated intracellular swelling in the development of altitude illness.⁸

A further major finding is that although hypoxia per se only affected supratentorial structures (see Figure 2), correlating T₂, mean diffusivity, and fractional anisotropy with headache score revealed significant alterations in water mobility in additional subtentorially located cerebral structures. For example, significant relationships were observed bilaterally within the cerebellum, corticospinal tract, ponds, and thalamic radiation with high-altitude headache intensity (see Figures 4 and 5). These central structures are intuitively linked to altitude illness by way of the trigeminocervical complex and the observed relative increase in sympathetic activity, catecholamines, and low-frequency to high-frequency ratio within the R-R variability power spectrum.^28,29 Furthermore, ataxia is a common feature of more severe altitude illness (e.g., high-altitude cerebral edema) that suggests hypoxia may affect neuronal traffic within the cerebellum. Thus, these data are the first to confirm altered brain water mobility in the brain stem and cerebellum during periods of acute hypoxia when symptoms of high-altitude headache are present. However, given our correlative findings, future work is needed to distinguish whether high-altitude headache is caused by the observed global changes in brain water, regional changes within the brainstem, or an interactive combination of supratentorial and subtentorial structures.

Limitations of the current investigation include the examination of only white matter structures. Indeed, neurons exhibit a substantial energy turnover compared with glial and oligodendrocytes and thus may be more susceptible to periods of hypoxia.² Furthermore, the current investigation only identifies the short-term effects of hypoxia and predominantly comprises individuals having moderate to severe high-altitude headache. Future research should aim to explore the differences in brain water mobility between individuals with high-altitude headache and those entirely resistant. In addition, corticoidsteroids such as dexamethasone are an established prophylactic treatment for acute mountain sickness but predominantly affect vasogenic edema and not intracellular swelling.³⁰ Critical experimental investigations using corticoidsteroids are therefore required to further test the cause and effect relationship between intracellular swelling and altitude illness. A final limitation is the possibility of a type I error when using multiple exploratory t-tests. However, given that our primary conservative analyses (TBSS and mROI) identified significant alterations in global and regional white matter water mobility, we deemed the possibility of a type I error an acceptable risk that could help to advance the literature by identifying novel susceptible brain regions. It is hoped that future investigators use the current data to formulate specific brain region hypothesis-driven analyses.

In conclusion, using a combination of whole-brain and region of interest-based analyses, this study is the first to report a significant widespread increase in white matter intracellular swelling without vasogenic edema during periods of acute hypoxia. Furthermore, these data suggest that changes in brain water mobility including areas such as the brain stem and cerebellum are related to the intensity of headache experienced in hypoxia.

DISCLOSURE/CONFLICT OF INTEREST

The authors declare no conflict of interest.

Footnotes

Notes

Supplementary Information accompanies the paper on the Journal of Cerebral Blood Flow & Metabolism website ()

References

Roach

Hackett

. Frontiers of hypoxia research: acute mountain sickness. J Exp Biol 2001; 204: 3161–3170.

Hackett

Yarnell

Hill

Reynard

Heit

McCormick

. High-altitude cerebral edema evaluated with magnetic resonance imaging: clinical correlation and pathophysiology. JAMA 1998; 280: 1920–1925.

Kallenberg

Bailey

Christ

Mohr

Roukens

Menold

. Magnetic resonance imaging evidence of cytotoxic cerebral edema in acute mountain sickness. J Cereb Blood Flow Metab 2007; 27: 1064–1071.

Schoonman

Sándor

Nirkko

Lange

Jaermann

Dydak

. Hypoxia-induced acute mountain sickness is associated with intracellular cerebral edema: a 3T magnetic resonance imaging study. J Cereb Blood Flow Metab 2007; 28: 198–206.

Bailey

Bärtsch

Knauth

Baumgartner

. Emerging concepts in acute mountain sickness and high-altitude cerebral edema: from the molecular to the morphological. Cell Mol Life Sci 2009; 66: 3583–3594.

Fischer

Vollmar

Thiere

Born

Leitl

Pfluger

. No evidence of cerebral edema in severe acute mountain sickness. Cephalalgia 2004; 24: 66–71.

Mórocz

Zientara

Gudbjartsson

Muza

Lyons

Rock

. Volumetric quantification of brain swelling after hypobaric hypoxia exposure. Exp Neurol 2001; 168: 96–104.

Hunt

Jr Theilmann

Smith

Scadeng

Dubowitz

. Cerebral diffusion and T2: MRI predictors of acute mountain sickness during sustained high-altitude hypoxia. J Cereb Blood Flow Metab 2012; 33: 372–380.

Assaf

Pasternak

. Diffusion tensor imaging (DTI)-based white matter mapping in brain research: a review. J Mol Neurosci 2007; 34: 51–61.

10.

Bailey

Roukens

Knauth

Kallenberg

Christ

Mohr

. Free radical-mediated damage to barrier function is not associated with altered brain morphology in high-altitude headache. J Cereb Blood Flow Metab 2005; 26: 99–111.

11.

Bailey

Taudorf

Berg

Lundby

McEneny

Young

. Increased cerebral output of free radicals during hypoxia: implications for acute mountain sickness? Am J Physiol Regul Integr Comp Physiol 2009; 297: R1283–R1292.

12.

Roach

Biärtsch

Oelz

Hackett

. The Lake Louise acute mountain sickness scoring system. In: Sutton

Houston

Coates

. (eds) Hypoxia and Molecular Medicine. Queen City Press: Burlington, VT, 1993,pp 272–274.

13.

Harris

Wenzel

Thomas

. High altitude headache: efficacy of acetaminophen vs. ibuprofen in a randomized, controlled trial. J Emerg Med 2003; 24: 383–387.

14.

Lawley

Oliver

Mullins

Morris

Junglee

Jelleyman

. Optic nerve sheath diameter is not related to high altitude headache: a randomized controlled trial. High Alt Med Biol 2012; 13: 193–199.

15.

Smith

Jenkinson

Johansen-Berg

Rueckert

Nichols

Mackay

. Tract-based spatial statistics: voxelwise analysis of multi-subject diffusion data. Neuroimage 2006; 31: 1487–1505.

16.

Smith

Jenkinson

Woolrich

Beckmann

Behrens

Johansen-Berg

. Advances in functional and structural MR image analysis and implementation as FSL. Neuroimage 2004; 23(Suppl 1): S208–S219.

17.

Zhang

Lin

Sun

Huang

Wang

. Compromised white matter microstructural integrity after mountain climbing: evidence from diffusion tensor imaging. High Alt Med Biol 2012; 13: 118–125.

18.

Rostrup

Larsson

Born

Knudsen

Paulson

. Changes in BOLD and ADC weighted imaging in acute hypoxia during sea-level and altitude adapted states. Neuroimage 2005; 28: 947–955.

19.

Zhang

Yan

Shi

Gong

Weng

Liu

. Structural modifications of the brain in acclimatization to high-altitude. PLoS ONE 2012; 5: e11449.

20.

Myerson

Loman

. Internal jugular venous pressure in man: its relationship to cerebrospinal fluid and carotid arterial pressures. Arch NeurPsych 1932; 27: 836–846.

21.

Loman

Myerson

Goldman

. Effects of alterations in posture on the cerebrospinal fluid pressure. Arch NeurPsych 1935; 33: 1279–1295.

22.

Basnyat

. High altitude cerebral and pulmonary edema. Travel Med Infect Dis 2005; 3: 199–211.

23.

Dyer

Hopkins

Perthen

Buxton

Dubowitz

. Regional cerebral blood flow during acute hypoxia in individuals susceptible to acute mountain sickness. J Respir Physiol Neurobiol 2008; 160: 267–276.

24.

Willie

Macleod

Shaw

Smith

Tzeng

Eves

. Regional brain blood flow in man during acute changes in arterial blood gases. J Physiol 2012; 590: 3261–3275.

25.

Moller

Paulson

Hornbein

Colier

Paulson

Roach

. Unchanged cerebral blood flow and oxidative metabolism after acclimatization to high altitude. J Cereb Blood Flow Metab 2002; 22: 118–126.

26.

Smith

Krizay

Guo

Shin

Scadeng

Dubowitz

. Sustained high-altitude hypoxia increases cerebral oxygen metabolism. J Appl Physiol 2013; 114: 11–18.

27.

Mandell

Han

Poublanc

Crawley

Kassner

Fisher

. Selective reduction of blood flow to white matter during hypercapnia corresponds with leukoaraiosis. Stroke 2008; 39: 1993–1998.

28.

Loeppky

Icenogle

Maes

Riboni

Scotto

Roach

. Body temperature, autonomic responses, and acute mountain sickness. High Alt Med Biol 2003; 4: 367–373.

29.

Duplain

Vollenweider

Delabays

Nicod

Bartsch

Scherrer

. Augmented sympathetic activation during short-term hypoxia and high-altitude exposure in subjects susceptible to high-altitude pulmonary edema. Circulation 1999; 99: 1713–1718.

30.

Fishman

. Brain edema. N Engl J Med 1975; 293: 706–711.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.03 MB

0.15 MB

0.14 MB

0.13 MB

0.12 MB