Hippocampal metabolites in asthma and their implications for cognitive function.

Emerging research indicates that individuals with asthma have an increased risk of cognitive impairment, yet the associations of asthma with neural correlates of memory remain relatively unknown. The hippocampus is the predominant neural structure involved in memory, and alterations in the hippocampal metabolic profile are observed in individuals with mild cognitive impairment. We therefore hypothesized that individuals with asthma may have altered hippocampal metabolites compared to healthy controls. Structural magnetic resonance imaging (sMRI) and proton magnetic resonance spectroscopy (1H-MRS) were used to compare hippocampal volume and metabolites of otherwise healthy adults with and without asthma (N = 40), and to study the association of these measures with cognitive function and asthma-related variables. Participants underwent 3-Tesla sMRI and 1H-MRS, with the volume of interest placed in the left hippocampus to measure levels of N-acetylaspartate (NAA), glutamate (Glu), creatine (Cr), and myo-inositol (MI), as indicators of neuronal viability, cellular activity, cellular energy reserve, as well as glial activation. Individuals with asthma had lower hippocampal NAA compared to healthy controls. For all participants, poorer cognitive function was associated with reduced NAA and Glu. For individuals with asthma, poorer cognitive function was associated with reduced disease control. Additionally, short-acting rescue bronchodilator use was associated with significantly lower NAA, and Glu, whereas inhaled corticosteroid use was related to significantly higher Cr and in tendency higher NAA and Glu. All findings controlled for left hippocampal volume, which was not different between groups. These findings highlight that asthma and/or its treatment may affect hippocampal chemistry. It is possible that the observed reductions in hippocampal metabolites in younger individuals with asthma may precede cognitive and hippocampal structural deficits observed in older individuals with asthma.

Introduction

Emerging research indicates that individuals with asthma across the life span suffer from higher rates of cognitive and memory impairment compared to the healthy population (Irani et al., 2017); however, biological correlates of these behavioral deficits remain unknown. A recent meta-analysis synthesizing over 4148 participants, found a medium effect sized relation between asthma and reduced cognitive function (Irani et al., 2017), and in a large community sample of individuals over 55 years, individuals with asthma had a 78% increased risk for the presence of mild cognitive impairment (Caldera-Alvarado et al., 2013). Despite an accumulation of studies indicating the risk of cognitive deficits in asthma, the impact of asthma on neural regions involved in memory and their chemistry remains relatively unexplored. Central nervous system (CNS) processes in asthma and dyspnea have recently attracted attention (Binks et al., 2014; Von Leupoldt et al., 2009a; Peiffer et al., 2008; Evans, 2010; Rosenkranz et al., 2012; Pattinson, 2015; Raux et al., 2013); however, any influence of asthma on neural regions involved in memory, has only been explored in one study, observing smaller hippocampal volumes in middle aged individuals with asthma (Carlson et al., 2017). The influence of asthma on neural chemistry has, to the best of our knowledge, not been explored.

The primary neural region involved in memory is the hippocampus, a structure integral for encoding episodic memory and memory consolidation (Dudek et al., 2016). Changes in hippocampal volume and chemistry often coincide with reduced cognitive performance (Brown et al., 2004; Kantarci et al., 2002); however, no studies have examined the hippocampal chemistry of patients with asthma.

Neuroimaging technology has advanced from purely structural insights and functional imaging to non-invasive identification of chemical metabolite markers of CNS activity through magnetic resonance spectroscopy (1H-MRS), a technology offering valuable in-vivo neural information, which was once otherwise limited to animal models of memory (Maddock and Buonocore, 2012). MRS can capture the metabolite N-acetylaspartate (NAA), an indicator of neuronal density and integrity; myo-inositol (MI), a putative marker of glial activation; glutamate (Glu), an excitatory neurotransmitter; and creatine (Cr), which additionally includes phosphocreatine and is used as a marker for cellular energy and metabolism (Allaili et al., 2015). NAA is the second most concentrated molecule in the brain after Glu and provides a critical source of acetate for myelin lipid synthesis in oligodendrocytes. NAA facilitates energy metabolism in neuronal mitochondria, and is therefore used as a putative marker of neuronal number, health, and viability (Maddock and Buonocore, 2012). MI is preferentially concentrated in glial cells and is used as a putative marker of microglial activation. Neuroinflammation is characterized by an increase in MI, and an increase of MI in neural regions has been found to be associated with and precede dementia (Voevodskaya et al., 2016; Targosz-Gajniak et al., 2013). Glu is the most prevalent excitatory neurotransmitter and its role is critical for the establishment of long term potentiation (LTP), which strengthens neural connections leading to gains in working and declarative memory (Rupsing et al., 2011). Cr is a precursor of adenosine triphosphate and therefore an indicator of cellular energy metabolism and storage, thought to be a relatively stable neural metabolite in both health and disease (Maddock and Buonocore, 2012; Allaili et al., 2015). Abnormalities in these four neural metabolites (deficits, elevations, or changes in their ratios) are observed in those with cognitive impairment, and have even been found to predict disease progression and future cognitive decline (Tumati et al., 2013; Modrego et al., 2011; Voevodskaya et al., 2016; Targosz-Gajniak et al., 2013; Rupsing et al., 2011).

Hippocampal gray matter volume deficits and alterations in hippocampal metabolites are additionally observed in chronic pulmonary inflammatory disease states, including COPD (Shim et al., 2001; Li and Fei, 2013; Esser et al., 2016); however, to the best of our knowledge, there are no studies examining hippocampal metabolites in asthma. Patients with asthma may be uniquely influenced by natural sequelae of a chronic systemic inflammatory disease, moments of hypoxia, asthma medication, and/or behavioral influences commonly associated with asthma including diminished sleep quality. Outside of the context of asthma, these factors independently demonstrate detectable influences on the hippocampus and cognition in both human and animal studies (Guo et al., 2013; Takada et al., 2015; Brown, 2009; Elcombe et al., 2017). We therefore hypothesized that hippocampal metabolites would be altered in asthma compared to healthy controls, and further hypothesized that lower levels of NAA and Glu would be associated with poorer cognitive function. We additionally hypothesized that asthma medication would demonstrate influences on hippocampal metabolites. As prior research suggests that hippocampal volume can be reduced in asthma (Carlson et al., 2017), we also analysed hippocampal structure to exclude the possibility that differences in metabolite levels were secondary to hippocampal volume. Studies in patients receiving chronic corticosteroid treatment have found no significant differences between left and right hippocampal metabolites (Brown et al., 2004), consistent with reviews of hippocampal metabolites in mild cognitive impairment (Maddock and Buonocore, 2012). Individuals with greater cognitive deterioration have shown lower levels of NAA in the left temporal lobe (Maddock and Buonocore, 2012), and we therefore focused our analyses on the left hippocampus.

Methods and materials

Additional methodological details are provided in the online supplement.

Participants

Twenty patients with a physician's diagnosis of asthma were compared to twenty age- and gender-matched healthy controls. Exclusion criteria for all participants included: neurological or cardiovascular disease, any other chronic inflammatory disease, lung disease besides asthma, history of smoking, current major depressive episode, current or recent history (within one year) of substance related disorders including alcohol abuse, recreational drug use, history of any manic episode, and symptoms of schizophrenia, bipolar disorder, or psychosis. Individuals who used corticosteroids (oral and injected) within the past 3 months were additionally excluded. As a precaution for the scanning environment, participants with values of forced expiratory volume in one-second (FEV1) % predicted <70%, were excluded (National Heart, Lung, and Blood Institute (NHLBI), 2007). FEV1 is a clinical standard measure of mechanical lung function captured by spirometry.

General procedure

Participants completed questionnaires followed by a trained experimenter presentation of the Montreal Cognitive Assessment (MoCA) and spirometry. Within one week, administration of the Asthma Control Questionnaire (ACQ) and spirometry were repeated immediately prior to magnetic resonance imaging. Individuals with asthma were asked to refrain from taking rescue inhalers the day of either session and were encouraged to reschedule if needed. This study was approved by the University of Texas Southwestern Medical Center (STU 082011-038) and Southern Methodist University (2015-007-RITT) Institutional Review Boards. Written informed consent was obtained from all participants.

Magnetic resonance spectroscopy acquisition and analysis

MR assessments were carried out on a whole-body 3 T scanner (Philips Medical Systems, Best, The Netherlands), equipped with a whole-body coil for RF transmission and an 32-channel phased-array head coil for reception. Water-suppressed point-resolved spectroscopy (PRESS) data were acquired with TR = 2 s, TE = 112 ms, sweep width = 2.5 kHz, number of sampling points = 2048, and number of signal averages (NSA) = 256. Water suppression was obtained with a vendor-supplied four-pulse variable-flip-angle sub-sequence. First and second order shimming was carried out, using the fast automatic shimming technique by mapping along projections (FASTMAP). The RF carrier frequencies of the PRESS sequence were set at 2.5 ppm and were adjusted for B0 drifts in each excitation using a vendor-supplied tool (Frequency Stabilization). Unsuppressed water was acquired from the voxel for eddy current compensation and multi-channel combination. Spectral fitting was performed with LCModel software (Provencher, 1993), using in-house basis spectra that were computer simulated incorporating the PRESS volume localizing radio-frequency and gradient pulses. The basis set included NAA, Cr (creatine + phosphocreatine), Glu, glutamine, GABA, glycine, MI, lactate, glutathione, alanine, acetate, aspartate, ethanolamine, phosphorylethanolamine, scyllo-inositol, taurine, N-acetylaspartylglutamate, glucose, Cho (glycerophosphorylcholine + phosphorylcholine). The spectral fitting was conducted between 0.5 and 4.1 ppm. Cramér-Rao lower bounds (CRLB) were returned as percentage standard deviation by LCModel. Metabolites were estimated with reference to both water and tCr (Fig. 1).

Fig. 1

MRI voxel placement and 1H-MRS spectra from the hippocampus of a participant with asthma. Sagittal view demonstrates placement of the volume of interest in the left hippocampus. The area under the spectra represents the concentration of each metabolite in the identified tissue. Red line, baseline; blue line, in vivo data; green line, fit; light blue line, residual.

Abbreviations: Cr, creatine; NAA, N-acetylaspartate; MI, myo-inositol; Glu, glutamate. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Magnetic resonance imaging acquisition and analysis

Structural MRI was conducted on the 3 T scanner described above, immediately prior to MRS imaging. A high-resolution T1-weighted image called magnetization prepared rapid acquisition of gradient-echo (MPRAGE) sequence was collected with the following parameters: FOV = 256 × 256 × 160 mm3, TR/TE = 8.1 ms/3.7 ms, flip angle = 12°, 160 sagittal slices, voxel size = 1 × 1 × 1 mm3 and duration of 4 min. The segmentation of left hippocampal region was performed via FSL-FIRST software (Schoemaker et al., 2016). Briefly, the images were initially registered to the MNI152 standard space template in a two-stage process, where the first stage was a whole-brain and the second stage was a subcortical-weighted 12 degrees of freedom linear fit. The registration was visually checked for each participant. The inverse of this transformation was applied to the segmentation model in order to bring it into the native space of the original (non-interpolated) T1-weighted image, where the segmentation was performed. In order to account for different brain sizes, we normalized the volume of the left hippocampus by the total intracranial volume (ICV). The ICV was calculated via FSL's BET and FLIRT routines.

Spirometry

FEV1 was measured with a handheld spirometer (CareFusion Jaeger/Toennies AM-2) to capture the participants' best of three forced exhalations (National Heart, Lung, and Blood Institute (NHLBI), 2007).

Measures

Cognitive function was measured with the Montreal Cognitive Assessment (MoCA), a 10-minute global cognitive screening tool (Brown, 2009) designed to detect mild cognitive impairment (MCI) and assesses orientation, executive function, visuospatial skills, abstraction, language, memory, and attention. Scores range from 0 to 30, with those scoring above 26 extremely unlikely to meet MCI criteria (Nasreddine et al., 2005).

Asthma control was measured with the Asthma Control Test (ACT), a self-report measure assessing asthma control over the past month (Nathan et al., 2004), and the Asthma Control Questionnaire (ACQ), assessing asthma control over the previous week (Juniper et al., 1999). ACQ captures both self-report and lung function by spirometry (FEV1%predicted).

Statistical analyses

Independent two sample t-tests calculated between group (asthma vs. healthy) differences in hippocampal metabolites and volume. Non-parametric Mann-Whitney test was calculated for group differences in Glu, as it violated assumptions of normality. Pearson partial correlations calculated relations among: hippocampal metabolites, cognitive scores, medication use and asthma control, controlling for hippocampal volume. IBM SPSS (Version 24) was used for analyses, and all variables demonstrated properties of normality. As group differences were observed in Cr, total metabolite values in reference to water, rather than in reference to Cr, were used for analyses according to recommendations of Tumati et al. (2013). MRS data was low quality for 5 subjects (3 asthma, 2 healthy), whose MRS data was excluded from analyses. Questionnaire and behavioral measures were retained for the entire sample.

Results

Participants were on average 25 years old and had a high education level (most were recruited from a local university). There was neither a statistically significant difference between the groups on demographics, years of education, nor physical characteristics (Table 1). Of those with asthma, 13 had well-controlled asthma with ACT values >19, and only two participants reported a single lifetime occurrence of an asthma-related emergency room visit, none within the past year (Table 1). Asthma severity ranged from intermittent to severe, based on National Asthma Education and Prevention Program guidelines (National Heart, Lung and Blood Institute, 2007) (Table 1). Individual classifications of asthma severity, asthma control scores, and medication prescriptions are provided in the supplement.

Table 1

Participant demographics and characteristics.

	Asthma n = 20			Healthy Controls n = 20			p
	Mean	SD	Min–Max	Mean	SD	Min–Max
Age (years)	25.3	8.9	18–44	25.10	8.88	18–45	0.958
Gender (male)	50%			50%			1.00
Education (years)	14.9	2.5	12–23	15.10	3.06	12–25	0.800
Height (m)	1.7	0.1	1.5–1.9	1.70	0.09	1.52–1.85	0.893
Weight (kg)	71.3	17.5	45.4–127.0	76.04	17.21	53.5–117.9	0.391
BMI (kg/m2)	24.5	4.8	17.7–38.0	26.39	5.38	18.5–37.7	0.261
MoCA (0−30)	27.1	2.5	21–30	28.05	2.01	23–30	0.190
FEV1 (% predicted)	95.0	10.5	77.0–116.0	100.3	13.3	80.0–130.0	0.366
L hippocampal volume	2.4e3	3.2e4	1.5e3–2.8e3	2.4e3	3.1e4	1.7e3–2.9e3	0.677
Asthma duration (years)	17.10	8.03	7–44
Lifetime asthma-related ER visits	0.10	0.31	0–1
ACT (5–25)	23.30	3.76	12–25
ACQ (0–6)	0.99	0.55	0.14–2.14
Asthma severitya (%)
Intermittent	22
Mild (persistent)	44
Moderate (persistent)	22
Severe (persistent)	11
Race (%)							0.617
White/Caucasian	60			45
Asian	5			0
Native American	0			0
African American	5			30
Other	10			5
Ethnicity (%)							0.937
Hispanic	20			20

Abbreviations: BMI, body mass index; FEV1, forced expiratory volume in 1 s; MoCA, Montreal Cognitive Assessment; ACT, Asthma Control Test; ACQ, Asthma Control Questionnaire.

Note, independent samples (two-tailed) t-test found no statistically significant difference between groups on any variables. Ranges of questionnaire scores are provided in parentheses after each measure. ACT is scored with higher values indicating better control and ACQ is scored with lower values indicating better control.

Asthma severity was calculated for participants on short-term medication and for those on maintenance medication whose asthma was well controlled according to NIH/NAEPP (2007) severity and medication step therapy guidelines. Two individuals who were on maintenance medication and did not have well-controlled asthma according to the ACT, were therefore excluded from severity classification.

Differences between asthma and healthy groups

Individuals with asthma had lower levels of both NAA, t (33) = −3.23, p = 0.003 (d = 1.12) and Cr, t (33) = −3.31, p = 0.002 (d = 1.15) compared to controls (Table 2), and demonstrated a nonsignificant trend in that direction with Glu, U = 204.0, p = 0.096 (d = 0.53) (Fig. 2). There were no statistically significant group differences in hippocampal volume, t (38) = 0.42, p = 0.677, MI, t (33) = −0.65, p = 0.521, or global cognitive function scores, t (38) = −1.34, p = 0.19.

Table 2

Total left hippocampal metabolites in reference to water.

	Asthma (n = 17)			Healthy controls (n = 18)			p	r
	Mean	SD	Range	Mean	SD	Range
tCr⁎⁎	9.15	1.46	6.85–11.70	10.57	1.06	9.04–12.71	0.003	−0.49
tNAA⁎⁎	13.56	0.92	10.07–16.40	15.31	1.24	12.85–17.84	0.004	−0.49
tMI	11.20	2.16	7.35–14.90	11.72	2.56	6.95–15.93	0.521	−0.11
tGlu	9.52	2.01	6.19–12.65	10.33	0.75	9.00–11.41	0.096	−0.26

Independent samples (two-tailed) t-test between group (asthma vs. healthy controls). Non-parametric independent samples Mann-Whitney U test was used for tGlu.

Abbreviations: SD, standard deviation; r, effect size; tCr, total creatine; tNAA, total N-acetylaspartate; tMI, total myo-inositol; tGlu, total glutamate.

Indicates significant difference between groups p < 0.01.

Fig. 2

Group differences on total values of left hippocampal metabolites. Abbreviations: Cr, creatine; NAA, N-acetylaspartate; MI, myo-inositol; Glu, glutamate.

Hippocampal metabolites and cognitive function

Controlling for hippocampal volume, MoCA scores were significantly correlated with total Glu (r = 0.35, p = 0.038) and demonstrated a trend in this direction with NAA (r = 0.29, p = 0.092) for all participants, suggesting that those with higher levels of these metabolites, in particular Glu, performed better on a measure of global cognitive function (Table 3, Fig. 3). In individuals with asthma only, the correlations controlling for hippocampal volume followed a similar pattern; however, they were not statistically significant (Online Supplement).

Table 3

Correlations among hippocampal metabolites, global cognitive function, and asthma control.

Variables	1	2	3	4	5
1. tCr	–
2. tNAA	0.58⁎⁎	–
3. tMI	0.33†	0.28†	–
4. tGlu	0.65⁎⁎	0.70⁎⁎	0.32†	–
5. MoCA	0.27	0.29†	0.16	0.35⁎	–
6. ACQ	−0.21	−0.14	−0.03	−0.31	−0.46⁎

Pearson partial correlations (two-tailed), controlling for hippocampal volume.

All individuals included in correlations with MoCA. Only individuals with asthma.

(n = 17) are included in correlations with ACQ. Pearson correlation (two-tailed) included for relation of ACQ and MoCA.

Abbreviations: ACQ, Asthma Control Questionnaire, higher scores indicate poorer control; MoCA, Montreal Cognitive Assessment, higher scores indicate better cognitive function; tCr, total creatine; tNAA, total N-acetylaspartate; tMI, total myo-inositol; tGlu, total glutamate.

Indicates p < 0.05.

Indicates p < 0.01.

0.05 < p < 0.10.

Fig. 3

Residual plots of left hippocampal metabolites and global cognitive function scores captured by the Montreal Cognitive Assessment, controlling for hippocampal volume in the entire sample. Statistics for Pearson partial correlations (two-tailed) displayed, controlling for left hippocampal volume. Abbreviations: Cr, creatine; NAA, N-acetylaspartate; MI, myo-inositol; Glu, glutamate.

Hippocampal metabolites and asthma medication

Frequency of short-acting bronchodilator use in the past week was negatively correlated with total NAA (r = −0.56, p = 0.020) and Glu (r = −0.58, p = 0.014), controlling for hippocampal volume, and demonstrated a nonsignificant correlation in that direction with Cr (r = −0.39, p = 0.138), indicating that more frequent use of rescue inhaler was related to lower levels of hippocampal metabolites (Table 4, Fig. 4). On the other hand, inhaled corticosteroid use (yes/no) was positively correlated with total Cr (r = 0.70, p = 0.002) and demonstrated a positive trend with NAA (r = 0.43, p = 0.082) and Glu (r = 0.40, p = 0.113), indicating that those on inhaled corticosteroids, clinically recommended for long-term asthma control, had higher levels of Cr (Table 4).

Table 4

Correlations among hippocampal metabolites and asthma medication use.

Variables	1	2	3	4	5
1. tCr	–
2. tNAA	0.59⁎	–
3. tMI	0.33	0.42†	–
4. tGlu	0.73⁎⁎	0.76⁎⁎	0.57⁎	–
5. Inhaled corticosteroid use (yes/no)	0.70⁎⁎	0.43†	0.38	0.40	–
6. Short-acting bronchodilator use (past week frequency)a	−0.39	−0.56⁎	−0.12	−0.58⁎	−0.19

Pearson partial correlations (two-tailed), controlling for hippocampal volume.

Abbreviations: tCr, total Creatine; tNAA, total N-acetylaspartate; tMI, total myo-inositol; tGlu, total glutamate.

p < 0.05.

p < 0.01.

0.05 < p < 0.10.

Item from the Asthma Control Questionnaire (ACQ).

Fig. 4

Residual plots of left hippocampal metabolites and frequency of short-acting bronchodilator use in the past week for the asthma group only, controlling for left hippocampal volume. Statistics for Pearson partial correlations (two-tailed) displayed, controlling for left hippocampal volume. Abbreviations: Cr, creatine; NAA, N-acetylaspartate; MI, myo-inositol; Glu, glutamate.

Hippocampal metabolites, asthma control, and cognitive function

Hippocampal metabolites were not significantly correlated with asthma control; however, asthma control for the week before the imaging assessment, measured by the ACQ, was correlated with the MoCA, (r = −0.46, p = 0.040), indicating that those with poorer asthma control had lower cognitive scores (Table 3, Fig. 5). This finding was no longer significant when controlling for age.

Fig. 5

Relations of Asthma Control Questionnaire (ACQ) and MoCA (Montreal Cognitive Assessment) scores. Pearson correlation (two-tailed) displayed.

Additional results of metabolite/Cr ratios and exploratory correlations among metabolites, disease duration and asthma-related nocturnal awakenings are included in the Online Supplement.

Discussion

Key findings

The present work demonstrates that in younger individuals with asthma, there are substantial changes in the hippocampal chemical profile. Even in the absence of cognitive impairment, those with asthma had reductions in both hippocampal NAA, a marker of neuronal integrity and Cr, a marker of cellular energy metabolism. These metabolites, in particular Glu, were additionally related to lower cognitive performance, after controlling for hippocampal volume. There was no group difference in hippocampal volume in this cognitively healthy sample, which was on average 25 years younger than that of previous studies observing hippocampal volume deficits in asthma (Carlson et al., 2017). However, the changes already observed in hippocampal metabolites in asthma before the onset of gross structural changes, may inform our understanding of the biological contributions to cognitive impairment observed in individuals with asthma over the lifespan.

Hippocampal metabolites in asthma

Our findings of reduced hippocampal NAA, already present in younger patients with asthma compared to age and gender matched healthy controls, complement previous observations of hippocampal volume deficits (Carlson et al., 2017) and increased risks of mild cognitive impairment (Caldera-Alvarado et al., 2013) in middle-aged individuals with asthma and affirm the influence of this chronic inflammatory disease not only on hippocampal structure, but also on its chemical composition. NAA is considered the most reliable MRS marker of brain dysfunction in mild cognitive impairment (Tumati et al., 2013), and lower levels of NAA are additionally observed to predict future cognitive decline (Modrego et al., 2011). Despite the absence of cognitive deficits in this younger sample, there are already differences in hippocampal NAA in individuals with asthma, which may precede future cognitive decline.

While the exact comparison of total NAA values observed in the present study to those previously reported is limited due to fewer studies reporting total NAA, differences in scanner strength, sample age, volume of interest size, and post processing methods, the reduction of hippocampal NAA in individuals with asthma demonstrated a medium effect size, consistent with the magnitude of change observed in mild cognitive impairment studies (Tumati et al., 2013). Compared to other disease groups who show reductions in hippocampal NAA compared to healthy controls, this sample demonstrated a smaller effect size compared to those with multiple sclerosis (Llufriu et al., 2014), schizophrenia (Maddock and Buonocore, 2012) and treatment resistant depression in youth (Lefebvre et al., 2017). The present findings suggest that there is a reduction in the hippocampal neuronal integrity of younger patients with asthma, which may be an important component to understand neural mechanisms contributing to the increased rate of cognitive deficits observed with asthma.

In addition to NAA, reductions in hippocampal Cr were observed in individuals with asthma. Cr is a marker of cellular energy thought to be stable neural metabolite in health and disease (Maddock and Buonocore, 2012; Allaili et al., 2015). As such, it has been used routinely as a reference for other metabolites. However, individuals with cognitive impairment have demonstrated reductions in hippocampal Cr, raising concerns for a potential Cr confound if metabolite/Cr ratios only are reported (for review see Tumati et al., 2013). As MRS technology has developed, a water reference can now be captured in the same scan, minimizing a potential confound of Cr, if it is influenced by disease process. The present findings support that concern and suggest the need for future studies to additionally capture total metabolic levels in asthma.

Despite similar levels of cognitive functioning across groups, the present findings may indicate pre-clinical influences of asthma control on global cognitive function. Poorer asthma control was correlated with lower cognitive scores, consistent with previous studies identifying variability in diurnal peak flow (used as a proxy for airway hyperreactivity, which is a hallmark of asthma) as the strongest predictor of cognitive function (O'Byrne et al., 2013). While it cannot be ruled out that poorer cognitive function leads to poorer asthma control with this cross-sectional study design, these first findings extend hypothesized influences of asthma on cognition beyond asthma severity (Irani et al., 2017).

While future longitudinal research is needed to identify specific mechanisms for cognitive impairment in asthma, the present findings of deficits in markers of hippocampal neuronal viability, cellular energy metabolism and their relations with poorer global cognitive scores provides first insight into neural processes that may contribute to poorer cognitive function observed in individuals with asthma (Irani et al., 2017).

Potential mechanisms

There are a number of potential explanations for the changes observed in these hippocampal metabolites.

First, inflammatory processes in asthma could affect hippocampal chemistry. Potential mechanisms include oxidative stress, which may influence both Th2 and Th1 immune responses and activate additional pro-inflammatory cytokines through NF-kB (Dozor, 2010). Active binding of peripheral cytokines to endothelial receptors may release additional mediators that impair blood-brain barrier integrity (Di Benedetto et al., 2017). Peripheral cytokines may also influence the CNS through stimulation of vagal sensory nerve or the sympathetic nervous system (Di Benedetto et al., 2017), which then induce central pro-inflammatory cytokine production, known to influence hippocampal chemistry (Barrientos et al., 2015).

Second, medication use may influence hippocampal metabolites. Systemic oral corticosteroid use has established dose-dependent influences on both poorer cognitive function and reduced hippocampal volume (Brown and Chandler, 2001; Brown, 2009), and individuals receiving chronic prednisone therapy have demonstrated reductions in hippocampal NAA/Cr (Brown et al., 2004). Participants in this study did not have recent exposure to systemic corticosteroids; however, use of such medications over a lifetime has demonstrated lasting influences on neurological and cognitive function (Shim et al., 2001). Although this study was not designed to elucidate the specific mechanisms of medication, exploratory analyses found evidence for positive effects of inhaled corticosteroids on Cr, NAA, and Glu. In contrast to high systemic doses of corticosteroids, local administration of this medication in small doses leads to better control of asthma (O'Byrne and Parameswaran, 2006) and therefore may have helped avoid some of the adverse CNS asthma sequelae. The additionally observed associations between more frequent rescue inhaler usage and lower levels of Cr, NAA, and Glu would be consistent with that interpretation. When individuals with asthma followed clinically recommended medication guidelines to achieve proper control (e.g. use of maintenance medication rather than relying on more frequent use of rescue inhalers; National Heart, Lung and Blood Institute, 2007), their hippocampal metabolites were more in line with those observed in the healthy sample. However, medication manipulation studies are needed to draw any further claims beyond this cross sectional association.

Third, hypoxia could be speculated to play a role, given its demonstrated influence on hippocampal structure, function and cognitive behavior in both animal and human studies (Takada et al., 2015). While mechanistically probable, prolonged moments of hypoxia are less frequent in physician monitored cases of controlled asthma and lifetime visits to the emergency room due to asthma exacerbation in this sample were infrequent (Table 1).

It is additionally possible that reductions in NAA and Cr may be influenced indirectly by behavioral factors common in chronic disease states, rather than the direct influence of asthma or its treatment. Sleep disruption, which is common in asthma, can influence hippocampal chemistry independently (Cross et al., 2013) and one of our supplemental findings, demonstrating an association between asthma-related night-time awakening and higher MI levels, points in that direction (Online Supplement). Equally important may be psychological factors, including anxiety and depression, which can additionally be comorbid with asthma and separately affect hippocampal chemistry (Brown et al., 2004). In other chronic respiratory disease states such as COPD, hippocampal activation during the anticipation of increased dyspnea is associated with behavioral variables including symptoms of dyspnea, anxiety, and reduced exercise capacity (Esser et al., 2017). This may be contrary to the present findings with asthma; however, it nonetheless indicates the importance of studying behavioral factors and the hippocampus in those with respiratory disease.

Further considerations

This is the first study, to the best of our knowledge, to test differences in the hippocampal metabolic profile of individuals with and without asthma, and further explore the relations between disease variables, cognition, and neural chemistry. This study not only provides first evidence of absolute differences in hippocampal neuronal integrity, but additionally highlights their importance, along with asthma control, in cognitive function.

Limitations of this study include a modest sample size, which may have led to lower power to detect differences in smaller MRS signals, such as Glu, which demonstrated reductions with a smaller effect size (Table 2). Nevertheless, there was evidence of robust group differences in NAA and Cr, with medium effect sizes. Neither older individuals, nor those with cognitive impairment were represented in this sample and likely limited the strength of observable relations among asthma, hippocampal chemistry and cognition. However, our restricted range established a more conservative estimate of such influences and highlights the fact that reductions in the hippocampal neuronal integrity are already observed at a younger age and may precede cognitive impairments observed in older adults with asthma. The brevity and global nature of the MoCA likely limited the measure's sensitivity to detect uniform statistically significant relations with metabolites of a neural region primarily involved in memory. However, the consistent pattern of relations across multiple metabolites contributes to increased confidence in their relations to cognition. The associations of the MoCA and asthma control may indicate that additional cognitive domains, such as exectuive function, are more associated with asthma control, a finding that is consistent with studies observing poorer academic performance observed in childhood asthma (Liberty et al., 2010). As such, a more comprehensive cognitive assessment, specific to memory, may yield stronger relations between metabolites and cognitive performance and the association between asthma control and MoCA scores may indicate asthma control may be related to multiple domains of cognitive function, rather than memory alone. Although we observed a relation between use of inhaled corticosteroids and hippocampal metabolites in this sample, medication use was recorded here as a binary variable to capture the comparison between corticosteroid-naïve participants and those who were prescribed inhaled corticosteroids. A more invested measurement of medication adherence and dosage over time using electronic monitoring was beyond the scope and means of our study, but will be important to capture in future studies of medication influences on hippocampal chemistry.

Despite these limitations, our findings demonstrate, for the first time, changes in hippocampal chemical profile of individuals with asthma, and their relations with cognitive performance. Future studies will require: replication with larger samples; inclusion of both left and right hippocampal metabolites to assess for potential lateralization (Shipton et al., 2014); extensions to additional neural regions relevant to cognition, asthma, and dyspnea (e.g. prefrontal cortex, insula, cingulate cortex, periaqueductal gray (Takada et al., 2015; Von Leupoldt et al., 2009b); use of diffusion tensor imaging to determine integrity of white matter tracts in asthma (Dodd et al., 2012); markers of peripheral immune function; and comprehensive neuropsychological assessment exploring specific components of memory. Studies with a longitudinal perspective could further explore potential moderating variables including: medication use, disease duration, asthma control, and sleep. Studies that are designed to test longitudinal mediation models, particularly for medication usage, will be important to include in future research.

The present study indicates, for the first time, that hippocampal neuronal integrity is reduced in asthma compared to healthy controls, providing insight into neural mechanisms, which may ultimately influence cognitive deficits observed in individuals with asthma. These findings provide additional evidence to support the recommendations (Irani et al., 2017) to screen for presence of cognitive impairment in individuals with asthma, particularly those who have poor disease control.