Brain and Lung Biomarker Responses to Hyperoxic Hypobaric Decompression
INTRODUCTION: Biomarker responses to intensive decompression indicate systemic proinflammatory responses and possible neurological stress. To further investigate responses, 12 additional brain and lung biomarkers were assayed. METHODS: A total of 15 healthy men (20 to 50 yr) undertook consecutive same-day ascents to 25,000 ft (7620 m), following denitrogenation, breathing 100% oxygen. Venous blood was sampled at baseline (T0), after the second ascent (T8), and next morning (T24). Soluble protein markers of brain and lung insult were analyzed by enzyme-linked immunosorbent assay with plasma microparticles quantified using flow cytometry. RESULTS: Levels of monocyte chemoattractant protein-1 and high mobility group box protein 1 were elevated at T8, by 36% and 16%, respectively, before returning to baseline. Levels of soluble receptor for advanced glycation end products fell by 8%, recovering by T24. Brain-derived neurotrophic factor rose by 80% over baseline at T24. Monocyte microparticle levels rose by factors of 3.7 at T8 and 2.7 at T24 due to early and late responses in different subjects. Other biomarkers were unaffected or not detected consistently. DISCUSSION: The elevated biomarkers at T8 suggest a neuroinflammatory response, with later elevation of brain-derived neurotrophic factor at T24 indicating an ongoing neurotrophic response and incomplete recovery. A substantial increase at T8 in the ratio of high mobility group box protein 1 to soluble receptor for advanced glycation end products suggests this axis may mediate the systemic inflammatory response to decompression. The mechanism of neuroinflammation is unclear but elevation of monocyte microparticles and monocyte chemoattractant protein-1 imply a key role for activated monocytes and/or macrophages. Connolly DM, Madden LA, Edwards VC, Lee VM. Brain and lung biomarker responses to hyperoxic hypobaric decompression. Aerosp Med Hum Perform. 2024; 95(9):667–674.
Altitude chamber exposure of 15 men to repeated exertional hypobaria, representative of parachutist despatchers working at 25,000 ft (7620 m) altitude, enabled evaluation of the early pathophysiological responses to hyperoxic decompression stress.6 Immediately postexposure, fivefold elevation of the proinflammatory cytokine interleukin-6 (IL-6) was accompanied by increased circulating neutrophils (mean 70%) and monocytes (37%). IL-6 increments appeared associated with severity of venous gas emboli bubble load.5,6 Elevated levels of complement peptide C5a, C-reactive protein, and neutrophil gelatinase-associated lipocalin persisted 24 h postexposure, respectively indicating complement system activation, synthesis of acute phase proteins, and neutrophil activation, and supporting an acute proinflammatory innate immune response. Endothelial activation and disruption, implied by elevated C5a and C-reactive protein, was evidenced by increased total endothelial microparticles (MPs) and circulating levels of tissue factor (coagulation factor III, thromboplastin). Increased IL-6, C-reactive protein, C5a, and neutrophil and endothelial activation imply increased oxidative stress, supported by changes in circulating glutathione and superoxide dismutase.6
The hypobaric exposures were characterized by prolonged hyperoxia, breathing 100% oxygen, and by generation of heavy loads of venous gas emboli in most, but not all, subjects. Activation and disruption of the pulmonary endothelium may underpin the acute phase response after exposure to the highest partial pressure of oxygen and bombardment by venous gas emboli.5 These microbubbles obstruct alveolar capillaries and interact with cellular components and the vascular endothelium.2 They trigger cytokine release and leukocyte, endothelial, and complement activation, damaging surface glycocalyx and endothelial cells, compromising vascular integrity, increasing permeability, and facilitating further bubble adhesion.18 Release of endothelial MPs suggests endothelial dysfunction secondary to oxidative stress and interaction with bubbles.13 MPs may also act as foci for microbubble formation and mediate further endothelial injury at other sites.31,33
In this context, the nature of the association between nonhypoxic decompression stress and compromised brain white matter integrity remains opaque.15,16 The recent study afforded an opportunity to target biomarkers of potential brain insult. Levels of serum calcium-binding protein S100β and neuron-specific enolase were unaffected, respectively indicating gross preservation of blood-brain barrier (BBB) integrity and absence of neuronal cell body damage. However, there was a mean 10% elevation above baseline of circulating astrocyte-derived glial fibrillary acidic protein (GFAP) the day after exposure (P = 0.015). GFAP is a sensitive marker of subtle central nervous system (CNS) insult, with levels peaking at 20 h postinjury and correlating with injury severity and tissue oxygenation.1 However, baseline levels were highly variable between subjects, ranging by a factor of about 20-fold from lowest to highest, such that the absolute magnitudes of some increments were relatively slight and the significance of the finding was uncertain. In the event of brain insult, BBB permeability increases to avoid local neurotoxicity, with active efflux of glutamate by endothelial cells to maintain extracellular homeostasis.26 Serum glutamate levels were indeed elevated (mean ∼10%) immediately postexposure, recovering to baseline the following day, but this did not achieve statistical significance [F(2,28) = 2.69, P = 0.085].6
Together, these data may indicate cerebral white matter stress but remain inconclusive. To investigate this possibility further, and to also evaluate whether the systemic inflammatory response may have originated in the lung, we conducted specific follow-up assays using contemporary biomarkers, listed in Table I, focusing on indicators of CNS insult and alveolar injury. We are unaware of any previous research reporting acute responses of these biomarkers to hypobaric decompression stress. Our null hypotheses were that these would be unaffected by hyperoxic hypobaria.
METHODS
Subjects
The study adhered to the principles of the Declaration of Helsinki. The research was funded by the UK Ministry of Defence (MOD) and the experimental protocol was approved in advance by the MOD Research Ethics Committee, an independent body constituted and operated in accordance with national and international guidelines. The methodology is described in detail elsewhere, so it is summarized briefly here.5,6
Subjects were 15 healthy nonsmoker men from 20 to 50 yr of age, by chance comprising five under 30 yr (mean 24, range 20–28) and 10 over 40 yr (mean 46, range 41–50). Study entry screening excluded right-to-left vascular shunts and excess pre-existing white matter hyperintensities. The experiment comprised consecutive hypobaric chamber ascents to 25,000 ft (7620 m) for 60 and then 90 min, each following 60-min denitrogenation, breathing 100% oxygen throughout. Exposures were separated by an hour on the ground with subjects breathing air normally. Subjects simulated the duties of parachutist despatchers at ground level and altitude, so were intermittently moderately physically active at 25,000 ft.
Procedure
Decompressions were conducted in the high-performance hypobaric chamber of the Altitude Research Facility at MOD Boscombe Down, Wiltshire, United Kingdom. Baseline 20-mL venous blood samples were obtained pre-exposure (T0) by antecubital venepuncture before 08:30. A second postexposure (T8) sample was collected immediately following completion of the second ascent, approximately 8 h later. The following day, a third, midmorning (recovery) sample was collected around 24 h after commencing the first ascent (T24). Three instances of limb bend DCS curtailed experiments, all resolving with recompression and treatment with 100% oxygen for an hour at ground level. These subjects’ T8 samples were collected immediately upon completion of oxygen breathing. Citrated samples (for plasma) and ‘clotted’ samples (serum) were centrifuged, processed, labeled, packaged, and transported to the Clinical Laboratory at Defence Science and Technology Laboratory, Porton Down, Wiltshire, United Kingdom, where they were frozen at −80°C within 2 h of sampling for later batch transfer on dry ice to the University of Hull, Hull, United Kingdom.
Equipment (Assays)
Protein biomarkers were analyzed at the University of Hull by high-sensitivity enzyme-linked immunosorbent assay (ELISA) with MP counts analyzed by flow cytometry using antibodies as previously described.10 The 12 specific biomarkers investigated and their relevance are detailed in Table I, together with the associated test kit or antibody suppliers. These were Abcam (Cambridge, United Kingdom); Abbexa (Cambridge, United Kingdom); Becton Dickinson (Wokingham, United Kingdom); Bio-Techne for R&D Systems (Abingdon, United Kingdom); Life Technologies Ltd for Thermo Fisher Scientific (Paisley, United Kingdom); and Miltenyi Biotec (Woking, United Kingdom). ELISA test kits (and antibodies) were usually the most sensitive available. ELISAs were performed following the relevant manufacturers’ instructions and plates were read at the appropriate wavelength using a BioTek Synergy HT Microplate Reader running Gen5 software (BioTek now Agilent, Santa Clara, CA, United States). For fluorescence-activated cell sorting, plasma samples (25 μL) were incubated with appropriate antibodies (5 μL) for 30 min in the dark at room temperature. Phosphate-buffered saline (0.2 μm filtered, 150 μL) and AccuCheck counting beads (25 μL, PCB100, Invitrogen, Waltham, MA, United States) were added prior to MP quantification by flow cytometry. Samples were analyzed using a BD FACSCalibur flow cytometer running CELLQuest Pro software (BD Biosciences, San Jose, CA, United States). The MP gate was set as described previously using Megamix SSc beads (Biocytex, Marseille, France).3
Statistical Analysis
Hemoglobin and hematocrit levels remained stable throughout.6 Postexposure plasma and serum markers were corrected for minor fluctuations in plasma volume in accordance with convention.7 Consistent with our previous biomarker analyses, to deal with idiosyncratic results that were unrepresentative of the cohort, any values greater than three standard deviations from the cohort mean, at any sample time, were treated as outliers; all data for that individual were then removed from the dataset for that marker.6 On this basis, no more than one subject’s data required removal from any dataset and no subject had their data removed from more than one dataset. Datasets were examined for a normal distribution using the Shapiro-Wilk test (α = 0.05). Data transforms (log10, square root) were applied, where necessary, to achieve this. Inferential analysis employed one-way repeated measures analysis of variance for the factor “Sample Time” (T0, T8, T24), with the significance level set at P < 0.05. The assumption of sphericity was verified with Mauchly’s test and, if violated, Greenhouse-Geisser correction was applied automatically.
RESULTS
Of the ELISAs, relative to baseline, statistically significant postexposure changes were observed in levels of monocyte chemoattractant protein-1 (MCP-1), brain-derived neurotrophic factor (BDNF), high mobility group box protein 1 (HMGB1), and soluble (as opposed to membrane-bound) receptor for advanced glycation end products (RAGE). All of these datasets remained intact with the exception that one outlier was removed from the RAGE dataset. This individual had exceptionally high levels at all sample times, but nonetheless followed the same pattern of response observed with the remainder of the cohort. The remaining RAGE data were normally distributed. MCP-1 and BDNF data were normally distributed but required Greenhouse-Geisser correction. HMGB1 data were normalized by log10 transform. Fig. 1 shows the MCP-1, BDNF, HMGB1, and RAGE data together with the outcome of their respective repeated measures analysis of variance. At T8, MCP-1 and HMGB1 were significantly elevated while RAGE levels fell, all recovering to baseline by T24. On the other hand, BDNF levels became elevated at T24.
Citation: Aerospace Medicine and Human Performance 95, 9; 10.3357/AMHP.6391.2024
One outlier was excluded from the data for ubiquitin carboxy-terminal hydrolase L1, which were then normalized by log10 transform. Postexposure levels were not significantly affected [F(1.26,16.44) = 1.01, P = 0.35]. Levels of neurofilament light were undetectable in eight samples from five subjects at various sample times. After their exclusion, the data from the remaining 10 subjects were normally distributed and unaffected by the exposure [F(2,18) = 0.48, P = 0.63]. Brain creatine kinase was detectable in only 10 samples from 4 subjects and was not analyzed further. Similarly, tau protein was detectable in only 17 samples from 11 subjects at inconsistent sample times, and was not analyzed further. For completeness, Table II summarizes the data from all ELISAs.
Turning to the MP assays, CD14 data were normalized by log10 transform. A substantial increase in CD14 count was evident at T8, driven particularly by the responses of five subjects, all exhibiting recovery toward baseline values at T24. However, late elevation of mean CD14 count at T24 was sustained by delayed responses of another five subjects. Fig. 2 represents both cohort and individual CD14 data.
Citation: Aerospace Medicine and Human Performance 95, 9; 10.3357/AMHP.6391.2024
With respect to both CD324 and CD326 MPs, one subject generated extreme counts at T8 that were, respectively, more than 20 and 30 standard deviations from the T8 mean. After excluding his data, both datasets were normalized by square root transform. Counts rose slightly at T24, but no statistically significant or consistent response patterns were evident (Fig. 2). The glutamate aspartate transporter MP assays were technically challenging. Two samples were lost due to technical difficulty and two gross outliers were excluded, one at T0 and another at T8. In the remaining subjects, levels were undetectable or at very low (background) levels, precluding meaningful analysis.
Given their interrelationship, the ratio of HMGB1 to RAGE was calculated and analyzed at each sample time point, with summary data shown in Fig. 3. The ratio increased significantly at T8, by a mean of 30%, subsequently recovering to baseline at T24 [F(2,28) = 5.46, P < 0.01].
Citation: Aerospace Medicine and Human Performance 95, 9; 10.3357/AMHP.6391.2024
DISCUSSION
The basis for evaluating levels of soluble markers of neuroinflammation in peripheral blood is that inflammatory mediators released from glial activation cross the BBB and enter the systemic circulation, promoting proliferation of peripheral immune cells. On the other hand, cytokines originating from peripheral immune cells also enter the brain parenchyma and act on glial cells, while proinflammatory mediators also modify the permeability of the BBB. Hence, neuroinflammation may follow systemic inflammation.25,32
MCP-1 is a chemokine that promotes monocyte interaction with endothelium and, in the CNS, chemokines aid recruitment of resident immune cells, i.e., astrocytes and microglia, and infiltration of circulating monocytes.4 Physiological expression of MCP-1 is typically low but is subject to significant upregulation during inflammatory responses, including induction by cytokines or oxidative stress.23 Besides endothelial cells, monocytes, and macrophages, many cell types in the brain produce MCP-1, including astrocytes, microglia, and neurons,30 with receptors expressed on neurons, astrocytes, microglia, and the microvascular endothelium.21 The elevated MCP-1 observed at T8 suggests an early neuroinflammatory response.
In addition to the MCP-1 data, CD14 MP analysis offered further support for a key role for monocytes in mediating the pathophysiological response in the current study. CD14 glycoprotein functions as a receptor, mediating innate immune responses. CD14 MP are shed from activated monocytes by cell membrane vesiculation, with background levels reflecting the numbers of monocytes in the circulation.8 In the current study, monocyte levels rose significantly immediately postexposure, by a mean of 37% at T8 [F(2,28) = 14.84, P < 0.00005], before normalizing at T24. However, relative to baseline, mean CD14 MP levels approximately trebled at T8 and more than doubled at T24 (Fig. 2A), implying probable monocyte activation resulting in MP release, with response time varying between subjects (Fig. 2B). Proinflammatory effects of monocyte MPs include further monocyte activation and cytokine generation, endothelial activation with expression of adhesion molecules, modulation of coagulation pathways, and increased endothelial oxidative stress.8 In particular, activated monocytes also produce IL-6 and are likely to have contributed to the mean fivefold elevation of IL-6 at T8 in the current study [F(2,28) = 35.42, P < 0.00001].6 Targets for monocyte MPs include the alveolar endothelium, promoting recruitment of inflammatory leukocytes,19 and brain endothelial cells, inducing endothelial vesiculation and microvascular inflammation,28 while levels of circulating CD14 MP have also been correlated with severity of cerebrovascular ischemia.9
The elevation of soluble HMGB1 at T8 is also indicative of a neuroinflammatory response.12,20 HMGB1 is released actively by neurons and microglia as well as by activated dendritic cells and macrophages, the latter upregulated by complement peptide C5a.27 Notably, C5a levels were previously shown to be significantly increased postexposure in this study [F(2,28) = 4.13, P = 0.027].6 Extracellular HMGB1 binds RAGE, a pattern recognition receptor that is key to initiating and maintaining inflammatory responses, promoting production of cytokines, specifically including IL-6.22 Soluble RAGE is an anti-inflammatory decoy receptor, providing negative feedback by inhibiting membrane-bound RAGE signaling in competition with other ligands.11 Thus, the decrease in postexposure levels indicates a relative loss of anti-inflammatory protection, while the significantly increased HMGB1/RAGE ratio indicates a clear proinflammatory shift (Fig. 3).
The acute responses discussed above normalized by T24 (Fig. 1). However, levels of BDNF became significantly elevated at T24. BDNF is the most abundant protective neurotrophin in the brain, promoting neuronal survival, neurogenesis, and synaptic plasticity.29 It was included in the current study as a biomarker of recovery from cerebral insult, being upregulated to assist cellular recovery following CNS injury, specifically encompassing white matter damage.17 The elevation at T24 suggests induction to facilitate recovery following CNS insult and is consistent with the previous report of delayed elevation of GFAP, further implying that cerebral recovery from decompression stress remains incomplete at T24.6 This may have implications for recovery time between provocative occupational exposures.
Collectively, these biomarker data provide strong evidence of an acute neuroinflammatory response to hyperoxic decompression stress that settles quickly while generating a residual neurotrophic response. The latter does not resolve fully by the following day, consistent with the previous report of mildly elevated GFAP at T24. The magnitudes of these acute responses are modest and we interpret them as physiological (stress responses) rather than pathological. However, the extent to which these responses may be induced repeatedly (e.g., over a career) before possibly predisposing to pathological change remains unknown. The findings suggest that the earlier report of elevated serum glutamate by 10% at T8 could possibly originate from the CNS or be a more systematic stress response, and hence be relevant in the current context; its failure to reach statistical significance previously may be most likely to reflect a lack of study power for this marker.6
On the other hand, our chosen assays were among the most sensitive available. Where target protein (or microparticle) levels remained below the threshold of detection, they are assumed not to have been upregulated (or released) in response to hyperoxic decompression, or at least not consistently across the cohort. We do not recommend further pursuing those biomarkers in the context of decompression stress.
While an all-male test cohort of 15 participants was proportionate to the 95% male parachutist despatcher population, the lack of female representation is a clear limitation of this study. This was intentional when the study was originally designed in 2019, to avoid any possible confound due to sex differences in DCS risk, any influence of the ovarian cycle or effects of hormonal contraception. Following a detailed review that has informed an appropriate study design, a follow-on trial involving both men and women is scheduled to validate the findings reported here.
The composite stress imposed in this study encompasses repeated prolonged hyperoxia, decompression sufficient to generate heavy loads of venous gas emboli in most participants, and intermittent bouts of physical activity of sufficient intensity to increase microbubble generation and possibly systemic oxidative stress. So the second main limitation of this study is the inability to include controls to isolate the relative contributions of these stressors to the observed outcomes. Unfortunately, it is impractical to control for all possible permutations of these factors while conducting an applied study that imposes representative, multifactorial environmental stress to evaluate decompression sickness risk; further work would be necessary to evaluate isolated and pairwise influences of these stressors. However, the level of exertion imposed in the current study was modest by comparison with those associated with substantial cytokine responses, generally encompassing endurance events or exercise at maximal oxygen uptake to exhaustion. Similarly, while hyperoxia induces cerebral and pulmonary oxygen toxicity under hyperbaric conditions, and may cause pneumonitis under prolonged normobaric conditions, it is usually considered innocuous to healthy military aviators for the durations required at altitude. Indeed, the oxygen cascade is such that, beyond the lung, capillary and target tissue partial pressures are elevated only slightly above normoxic levels.24 Hence, the outcomes reported here may depend fundamentally on the composite nature of the decompression stress and its net consequences. Hyperoxic pulmonary sensitization appears inevitable. Three-way interaction of microbubbles, inflammatory leukocytes, and pulmonary endothelium will involve both neutrophils and monocytes, with complement activation driving the inflammatory response via the HMGB1-RAGE axis and generating IL-6.14 Physical activity may enhance oxidative stress and promote the consequent systemic inflammatory response that particularly impacts the CNS, experienced most sensitively in the subcortical white matter and possibly mediated by endothelial and monocyte-derived MP.13
Finally, this study provides some evidence of a transient CNS insult that is secondary to nonhypoxic decompression stress and does not resolve fully within 24 h. Some of the pathophysiological responses described here may underpin the association between altitude exposure and brain white matter change.15,16 Hyperoxia-induced cerebral vasoconstriction reduces tissue blood flow, while microvascular flow heterogeneity may lower tissue oxygen availability and increase shunting to the venous circulation.24 Under these circumstances, microvascular dysregulation may promote localized neuroinflammatory responses that could be most consequential in vulnerable vascular watershed regions, notably affecting the subcortical white matter, providing a putative link between repeated nonhypoxic hypobaria and focal white matter change over time. The novel biomarker data reported here warrant further evaluation and validation in the context of hyperoxia and nonhypoxic hypobaric decompression.
Soluble biomarker responses (mean ± SE) for: A) MCP-1; B) HMGB1; C) BDNF; and D) RAGE.
Microparticle results for: A) mean ± SE cohort CD14 counts; B) individual subject CD14 counts; C) mean ± SE CD324 counts; D) mean ± SD CD326 counts.
Cohort mean ± SE HMGB1/RAGE ratio at each sample time.
Contributor Notes
