Frontal Metabolites and Alzheimer’s Disease Biomarkers in Healthy Older Women and Women Diagnosed with Mild Cognitive Impairment

Abstract

Background:

Women account for two thirds of the prevalence and incidence of Alzheimer’s disease (AD) and mild cognitive impairment (MCI). Evidence suggest that sex may differently influence the expression of proteins amyloid-beta (Aβ_1–42) and tau, for which early detection is crucial in prevention of the disease.

Objective:

We investigated the effect of aging and cerebrospinal fluid (CSF) levels of Aβ_1–42 and tau on frontal metabolites measured with proton magnetic resonance spectroscopy (MRS) in a cohort of cognitively normal older women and women with MCI.

Methods:

3T single-voxel MRS was performed on the medial frontal cortex, using Point Resolved Spectroscopy (PRESS) and Mescher-Garwood Point Resolved Spectroscopy (MEGA-PRESS) in 120 women (age range 50–85). CSF samples of Aβ_1–42 and tau and scores of general cognition were also obtained.

Results:

Levels of frontal gamma aminobutyric acid (GABA+) were predicted by age, independently of disease and CSF biomarkers. Importantly, levels of GABA+ were reduced in MCI patients. Additionally, we found that levels of N-acetylaspartate relative to myo-inositol (tNAA/mI) predicted cognition in MCI patients only and were not related to CSF biomarkers.

Conclusion:

This study is the first to demonstrate a strong association between frontal GABA+ levels and neurological aging in a sample consisting exclusively of healthy older women with various levels of CSF tau and Aβ_1–42 and women with MCI. Importantly, our results show no correlation between CSF biomarkers and MRS metabolites in this sample.

Keywords

Aging cerebrospinal fluid biomarkers frontal cortex general cognition magnetic resonance spectroscopy mild cognitive impairment women’s health

INTRODUCTION

Women account for two thirds of the prevalence and incidence of Alzheimer’s disease (AD) and dementia [1]. Evidence suggest that risk of development, diagnosis, and management of AD may be impacted by sex and gender [2]. Furthermore, sex and gender could impact development of new therapies and clinical outcome [3]. Current explanations for the discrepancy in incidence suggest that women statistically live longer and older age is the primary risk-factor for developing mild cognitive impairment (MCI) and AD. Other specific biological variables include increased APOE ɛ4-related risk of developing AD [4, 5], interaction with sex hormone estrogen [6], and differential expression of proteins amyloid-beta (Aβ_1–42) and tau in women compared to men [7]. Both Aβ_1–42 and tau are pathological hallmarks of AD but are also risk-factors for development of AD in cognitively normal healthy older adults [8]. Early detection of elevated levels of Aβ_1–42 and tau in the healthy brain is crucial for prevention and management of the disease, as evidence suggests that accumulation of Aβ_1–42 may start well before appearance of symptoms, through parallel mechanisms of increased microglial response and neuroinflammation [9, 10].

Previous works have demonstrated that aging- and AD-related physiological factors such as neuroinflammation have an impact on the expression of neurochemical compounds measured with proton magnetic resonance spectroscopy (MRS). Indeed, studies have shown that MRS measurements of a wide range of metabolites and neurotransmitters [11] can characterize normal aging [12, 13], differentiate healthy older controls from AD/MCI patients [14 –16], and predict conversion of MCI to dementia [17 –19]. Additionally, prior works have also established relationships between the expression of metabolites and cognitive performance, in healthy and MCI samples [20, 21]. Such works have highlighted the importance of measuring levels of N-acetylaspartate (tNAA, N-acetylaspartate +N-acetylaspartylglutamate), a marker of neuronal health and basic neural metabolism, total choline (tCho; phosphocholine + glycerophosphocholine), a marker of myelination and lipidic membrane turnover, myo-inositol (mI), a potential marker of glial activation and inflammation, and total creatine (tCr; creatine + phosphocreatine) [22, 23]. Additionally, levels of glutamate + glutamine (Glx) approximating total glutamatergic activity, and levels of main inhibitory compound gamma-aminobutyric acid (GABA), can both be reliably estimated at 3T.

Importantly, studies in aging and AD have often highlighted the importance of gender-matching [3]. Since the incidence numbers in AD demonstrate that women are significantly more at risk, they represent a vulnerable population that should be investigated separately. One hypothesis is that since women are more at risk, they could also be differentially sensitive to early neurochemical changes ascribable to AD-related biomarkers Aβ_1–42 and tau. Previous works have indicated sex differences in neuroinflammation and cytokine signaling, based on interaction with different sex hormones [24, 25]. Sex hormones also directly interact with several neurotransmitter systems, including GABA signaling (for review, [26]), and a prior study has identified significant differences in levels of GABA and glutamate (Glu) as measured by MRS between women and men [27]. Taken together, the accounted differences between men and women in 1) incidence of pathology, 2) underlying mechanisms of inflammation, and 3) concentration levels of brain metabolites, suggest that it is warranted to investigate the neurochemistry of a cohort of women, in the context of normal and pathological brain aging.

In this paper, we investigated the neurochemical profile of a sample consisting of healthy older women with normal levels of Aβ_1–42 (biomarker negative; BM-), healthy older women with reduced levels of cerebrospinal fluid (CSF) Aβ_1–42 (biomarker positive; BM+), and women diagnosed with MCI. Specifically, using MRS of the medial frontal cortex, we investigated changes in GABA relative to levels of Aβ_1–42, p-tau, and t-tau, and aging. We also investigated compounds that are frequently associated with neuroinflammation in aging, that is myo-inositol, choline, and Glx (glutamate + glutamine). Finally, we investigated the relationship between these MRS and CSF variables and scores on an assessment of general cognition.

METHODS

Population

Data were acquired from 71 BM-, 37 BM+, and 12 MCI (total N = 120) (see Table 1). BM- and BM+ were recruited as community-dwelling older women without a diagnosis of MCI or AD. Subjects with a history of neurological or psychiatric disorder and/or brain injury, renal dysfunction, any MRI contraindications (e.g., pacemaker, metal implant) or currently battling life-threatening illness (e.g., cancer) were excluded from the study. MCI patients were recruited from the Goizueta Alzheimer’s Disease Research Center (ADRC) and were diagnosed by expert clinicians affiliated with the ADRC. We did not obtain additional information about menstrual cycle; the average age for menopause in the US is 49 years [28]. Informed and written consent was obtained from all subjects in accordance with procedures established through the Emory University Institutional Review Board. All subjects are part of the ongoing Emory Brain Imaging Project embedded in the Emory Healthy Brain Study [29]. Although men are also part of this study, too few had participated at time of analyses and were not included.

Table 1

Summary of demographics, cognitive measurements and CSF biomarkers obtained in each group (mean±SD)

		BM-	BM+	MCI
Total sample size		71	37	12
Age (y±SD)		65.2±6.2	66.2±5.9	67.6±10.6
Education (y±SD)		16.7±0.32	16.7±0.21	16.9±0.55
Cognitive measurement	MoCA	26.5±2.5	26.1±2.6	17.1±8.7
		N = 57	N = 33	N = 12
CSF biomarker measurement (pg/mL)	Aβ_1–42	1457.2±247.9	764.8±164.8	670.7±133.3
		N = 68	N = 37	N = 12
	p-tau		19.4±7.1	22.1±16.4	46.5±16.8
		N = 68	N = 37	N = 12
	t-tau		218.3±73.5	226.4±148.1	457.1±158.9
		N = 68	N = 37	N = 12

Magnetic resonance imaging (MRI)

Anatomical images were obtained on a 3T Siemens Prisma (Siemens AG, Erlangen, GER) scanner with a 32-channel coil. The high-resolution T1-weighted images were acquired with a T1 (MPRAGE) sequence with TR 2300 ms, TE 2.96 ms, TI 900 ms, slice thickness 1 mm, 208 slices, field of view 256×256 mm², flip angle 9 degrees, and isotropic resolution 1×1×1 mm.

Magnetic resonance spectroscopy (MRS)

MRS was performed with 1) a Point-Resolved Spectroscopy (PRESS) single-voxel sequence with a short-echo time (TE 20 ms, TR 2000 ms) with 100 averages, vector size 2048 points, flip angle 90 degrees, acquisition duration 1024 ms, acquisition bandwidth 2000 Hz, water suppression bandwidth 135 Hz, and 2) a Mescher-Garwood Point Resolved Spectroscopy (MEGA-PRESS) sequence (TE 68 ms, TR 2000 ms) with 148 averages, vector size 2048 complex data points, flip angle 90 degrees, acquisition bandwidth 2000 Hz, acquisition duration 1024 ms, editing pulse bandwidth 53 Hz, ON editing pulse 1.90 ppm, OFF editing pulse 7.50 ppm, water suppression bandwidth 135 Hz. B0 shim was adjusted using the FASTESTMAP algorithm as implemented in the CMRR Spectroscopy package C2P [30, 31], and frequency and eddy current corrections were performed prior to data processing. Data were obtained from a 3×3×3 cm³ voxel positioned in the medial frontal cortex (MFC), superior to the genu of the corpus callosum and aligned with the shape of the corpus callosum, encompassing parts of Brodmann’ areas 24, 32, and 6 (see Fig. 1). Although the frontal cortex is thought to be targeted later in the development of AD pathology, frontal areas, including MFC, have been found to be more predictive of cognitive deficits and early disturbance of activities in daily living [32, 33] and may be more sensitive to subclinical changes. It is possible that such subclinical changes could be attributable to changes in AD-related CSF biomarkers and frontal metabolites, a missing link in the literature since the frontal cortex is a less explored cerebral area in MRS studies in early AD and MCI patients.

Fig. 1

MRS voxel position over the medial frontal cortex (MFC) in the A) axial, B) sagittal, and C) coronal planes.

Spectra with poor water suppression, lipid contamination or movement artifact were excluded from further analyses. Concentration levels obtained from PRESS were estimated with LC Model [34] and MEGA-PRESS with Gannet 3.0 [35]. PRESS data were excluded when Cramer-Rao lower bounds (CRLBs) exceeded 10% and MEGA-PRESS data were excluded when normalized fitting residuals were below 10% to maintain a reliability standard in the estimation of metabolites. Levels of GABA were obtained from MEGA-PRESS, and all other metabolites from PRESS. Estimation of absolute GABA levels with Gannet 3.0 assumes a portion of macromolecules co-edited at 3 ppm and is thus noted GABA+. Although metabolites tNAA and Glx can be estimated from MEGA-PRESS data, they were obtained from the PRESS acquisition. Metabolites were analyzed in absolute levels and normalized to tCr. Normalization to tCr was utilized as internal control for changes in cellular metabolism within subject, and is frequently reported in MRS studies [36]. Additionally, the ratio tNAA/mI was computed in the present study since it has demonstrated sensitivity relative to AD pathology in prior studies in AD and MCI populations [14, 37]. Segmentation of the voxel was performed with SPM 12 on the anatomical images to obtain tissue fraction of gray matter (GM), white matter (WM), and CSF, and to perform partial volume correction. All MRS measurements were corrected for CSF partial volume effects using a two-compartment model prior to statistical analyses [38].

CSF biomarkers

Levels of biomarkers were obtained with lumbar CSF sampling following the procedure described in a multicentric study [39]. Concentration levels of Aβ_1–42, t-tau and phosphorylated tau₁₈₁ (p-tau) were obtained from one CSF sample per subject and estimated with immunoassay biochemical tests (ADRC Clinical Research Unit, Atlanta, USA) with a Roche Elecsys analytical platform (Roche Diagnostics, Basel, SWI). Classification of BM- and BM+ subjects was based on the ADNI threshold of CSF [Aβ_1–42] < 977 pg/mL determined on the same analytical platform [40]. In the present study, all MCI patients were categorized as BM+.

Neuropsychological assessments

General cognition was assessed with the Montreal Cognitive Assessment (MoCA) score [41].

Statistical analyses

In group-based comparisons, variables were entered in a one-way ANOVA with Group as main factor with 3 levels (BM-, BM+, MCI). Post-hoc pairwise comparisons were computed to assess differences between groups when the ANOVA proved significant. If normality of distribution and heterogeneity of variances were not verified, variables were entered in nonparametric Kruskal-Wallis test and Wilcoxon test for post-hoc pairwise comparisons.

Linear regression analyses were performed to evaluate correlation of metabolites (as predicting variable) with MoCA scores (as dependent variables), and correlation of CSF biomarkers (as predicting variables) with MRS metabolites and MoCA scores (as dependent variables). Results are reported significant at α= 0.05 after correcting for false discovery rate with the Hochberg-Benjamini procedure [42]. All regression models were controlled for age. Adjusted R² accounting for multiple predictors and covariates are reported. Group differences in regression slopes were compared by testing for interaction between variables. Analyses were conducted with R statistical packages (R project, Vienna, AUS).

RESULTS

Group comparisons of age

The ANOVA shows chronological age was not different in all three groups (BM- 66.8±4.83; BM+ 68.9±6.09; MCI 68.4±8.86) (see Table 1).

Group comparisons of CSF biomarkers

Levels of CSF Aβ_1–42 were significantly different (p < 0.001) between all group. This was expected to confirm the categorization of BM- and BM+ subjects based on Aβ_1–42. Levels of t-tau (p < 0.001) and p-tau (p < 0.001) were also significantly different between MCI and both BM- and BM+, but not between BM- and BM+. (CSF Aβ_1–42 BM- 1457.2±247.9; BM+ 764.8±164.8; MCI 670.7±133.3; t-tau BM- 218.3±73.5; BM+ 226.4±148.1; MCI 457.1±158.9; p-tau BM- 19.4±7.1; BM+ 22.1±16.4; MCI 46.5±16.8) (see Table 1).

Groups comparisons of MRS voxel tissue fraction

Results in BM-, BM+, and MCI groups show significant differences in GM (p = 0.017), WM (p = 0.028), and CSF (p < 0.001) tissue fraction. For GM fraction, post-hoc comparisons show that concentration of GM was significantly greater in BM- (p = 0.004) and BM+ (p = 0.042) compared to MCI. BM- and BM+ were not significantly different (p = 0.557) (GM fraction: BM- 0.439±0.039; BM+ 0.434±0.034; MCI 0.412±0.044). Concentration of WM was significantly greater in BM- (p = 0.012) and BM+ (p = 0.019) compared to MCI. BM- and BM+ were not significantly different (p = 0.811) (WM fraction: BM- 0.347±0.067; BM+ 0.349±0.052; MCI 0.309±0.063). Concentration of CSF was significantly lower in BM- (p < 0.001) and BM+ (p < 0.001) compared to MCI. BM- and BM+ were not significantly different (p = 0.865) (CSF fraction: BM- 0.214±0.073; BM+ 0.216±0.054; MCI 0.278±0.079).

Groups comparisons of metabolites concentration levels

Results show that levels of mI/tCr (ANOVA p =0.0044) were higher in MCI compared to BM- (p = 0.0029) and BM+ (p = 0.0218) but were not significantly different between BM- and BM+ (p =0.899) (mI/tCr: BM- 0.96±0.10; BM+ 0.97±0.05; MCI 1.06±0.05).

Results also show that levels of GABA+ (ANOVA p = 0.0045) were lower in MCI compared to BM- (p = 0.0027) and BM+ (p = 0.0251) but were not significantly different between BM- and BM+ (p = 0.595) (GABA+: BM- 0.98±0.14; BM+ 0.95±0.19; MCI 0.79±0.20).

Levels of Glx (ANOVA p = 0.0218) were higher in MCI compared to BM- (p = 0.0177) and BM+ (p = 0.045) but were not significantly different between BM- and BM+ (p = 0.996) (Glx: BM- 8.77±1.19; BM+ 8.79±1.05; MCI 9.92±1.75).

Results show no significant differences between groups in levels of tCr (p = 0.138), tNAA/tCr (p =0.544), tNAA/mI (p = 0.178), and Cho/tCr (p = 0.116) (see Table 2; see Fig. 3).

Table 2

Summary of MRS measurements obtained in each group (mean±SD)

		BM-	BM+	MCI
Total sample size		71	37	12
MRS measurement (A.U.)	tCr	7.25±0.50	7.92±0.76	7.73±0.54
		N = 59	N = 29	N = 12
	tCho	1.81±0.20	2.15±0.27	2.11±0.24
		N = 59	N = 29	N = 12
	tCho/tCr	0.26±0.03	0.27±0.03	1.81±0.02
		N = 59	N = 29	N = 12
	tNAA	10.17±1.35	9.87±1.19	10.39±1.91
		N = 59	N = 29	N = 12
	tNAA/tCr	1.29±0.15	1.34±0.18	1.26±0.13
		N = 59	N = 29	N = 12
	tNAA/mI	1.35±0.22	1.33±0.21	1.22±0.18
		N = 59	N = 26	N = 12
	Glx (Glu+Gln)	8.77±1.19	8.79±1.05	9.92±1.75
		N = 59	N = 29	N = 12
	mI	6.41±0.70	7.73±1.34	7.98±0.85
		N = 59	N = 29	N = 12
	mI/tCr	0.96±0.10	0.97±0.05	1.06±0.05
		N = 59	N = 29	N = 12
	GABA+	1.03±0.20	0.93±0.16	0.90±0.14
		N = 62	N = 35	N = 12
	GABA+/Cr	0.13±0.02	0.13±0.02	0.10±0.03
		N = 62	N = 35	N = 12

Fig. 3

Boxplots presenting groups comparisons of CSF biomarkers A) Aβ_1–42, B) p-tau, C) t-tau, and D) MoCA scores. Values for CSF biomarkers are given in pg/mL (^*p < 0.05; ^***p < 0.001).

Fig. 2

Sample MRS spectra for A) PRESS and B) MEGAPRESS acquisitions.

Fig. 5

Scatterplots presenting the relationship between levels of frontal GABA+ and Age in A) BM- subjects, B) BM+ subjects, and C) MCI patients. Levels of GABA+ are given in arbitrary units (A.U.) and age in years. Confidence interval is 95% for the regression lines.

Fig. 6

Scatterplots presenting the relationship between MoCA score and tNAA/mI ratio in A) BM- subjects, B) BM+ subjects, and C) MCI patients. Confidence interval is 95% for the regression lines.

Group comparisons of MoCA scores

The nonparametric ANOVA was significant (p < 0.001), showing that general cognition was lower in MCI compared to BM- (p < 0.001) and BM+ (p < 0.001), but not significantly different between BM- and BM+ (BM- 26.5±2.5; BM+ 26.1±2.6; MCI 17.1±8.7).

MRS metabolites as functions of AD-related biomarkers and age

The multivariate linear model estimated the relationship of MRS metabolites, as dependent variables, and levels of CSF Aβ_1–42, t-tau, and p-tau as predictors with age as covariate. Results show that CSF biomarkers could not predict any MRS metabolite in any group. However, results also show that levels of GABA+ were significantly predicted by age (p < 0.001, R² = 0.247, F = 7.935). Results of independent linear models within groups show that age significantly predicted GABA+ in BM- (p < 0.001, R² =0.207, β= –0.474), BM+ (p = 0.0036, R² = 0.206, β=–0.479), and MCI (p = 0.035, R² = 0.374, β= –0.666) (see Fig. 3).

MoCA as a function of MRS metabolites and AD-related biomarkers

MoCA was significantly predicted by levels of tNAA/mI and Group (p < 0.001, R² = 0.525, F = 17.37). Since the interaction between tNAA/mI and Group was significant (p < 0.001), we proceeded to stratified analyses within each group. The analysis in the MCI group only proved significant, suggesting that higher MoCA score is predicted by higher levels of tNAA/mI (p = 0.038, R² = 0.328, β= 0.629) (see Fig. 4). MoCA was not significantly predicted by any other MRS metabolite. Finally, MoCA was not correlated with levels of CSF Aβ_1–42, t-tau, and p-tau in any group.

Fig. 4

Boxplots presenting groups comparisons of MRS metabolites A) GABA+, B) mI/tCr, C) Glx (Glu + Gln), D) tNAA/tCr, E) tCr, and F) tNAA/mI. Values are given in arbitrary units (A.U.) (^*p < 0.05 ^***p < 0.001).

DISCUSSION

The prevalence of AD and dementia is significantly higher among women. Multiple sex- and gender-specific factors could explain such a difference in the pathophysiology of AD [6, 43]. Studies have demonstrated the importance of proteins Aβ_1–42 and tau in the early pathology of AD, albeit their respective physiological roles are still not fully understood. In this study, we sought to investigate the relationship between MRS metabolites and levels of CSF Aβ_1–42, p-tau, and t-tau in a sample of older women using multiple MRS sequences in the medial frontal cortex. Additionally, we obtained MoCA scores and observed relationships between general cognition and MRS metabolites and AD-related CSF biomarkers.

A primary finding of this study is that age and MCI pathology exhibit independent detrimental effects on frontal levels of GABA+ in this sample of healthy older women and MCI patients. First, our results show that levels of frontal GABA+ significantly decrease with age and this relationship remains significant in all groups in independent analyses. This is in line with previous studies that have documented a decrease in frontal and posterior GABA concentrations with healthy aging [13, 20]. A recent study has also shown that MCI patients had lower prefrontal GABA levels, which were positively correlated with memory performance [44]. Other results have shown reduced levels of GABA in the posterior cingulate cortex of MCI patients [45, 46]. Our results seem to show that this age-related decrease of GABA is also present in the medial frontal areas.

Secondly, our results show that levels of GABA+ were significantly lower in MCI patients compared to both BM- and BM+ subjects but were not significantly different between BM- and BM+. This suggests that MCI-related pathology, not elevated levels of CSF Aβ_1–42 or tau, predicts lower levels of GABA+. Indeed, in BM+ subjects, levels of Aβ_1–42 and tau were not correlated to GABA+. The study from Riese and colleagues (2015) had shown that decreased levels of GABA were not correlated to levels of PET Aβ_1–42 nor APOE ɛ4 expression in MCI patients, suggesting that levels of GABA may reflect cognitive decline rather than physiological progression of preclinical AD. To the best of our knowledge, the current study is the first to demonstrate a strong association between frontal GABA+ levels and neurological aging in a sample of healthy older women and women with MCI.

Another notable finding of this study is that levels of frontal tNAA/mI were correlated to general cognition in MCI patients only. Specifically, lower levels of tNAA/mI predicted poorer scores on the MoCA, and that relationship was not significant in either BM- or BM+. Importantly, levels of tNAA/mI were not significantly lower in the MCI group. The tNAA/mI ratio has been demonstrated as an efficient MRS marker for cognitive decline: studies suggested that posterior tNAA/mI levels can predict general cognition in MCI patients [37] and posterior cingulate tNAA/mI levels predicted appearance of MCI symptoms in healthy older populations [16]. Interestingly, previous findings have also shown that lower levels of tNAA/mI in the posterior cingulate were associated with increased PET Aβ_1–42 accumulation over time in cognitively normal older adults [14]. In our sample, CSF levels of Aβ_1–42 were not associated with tNAA/mI or other metabolites, or MoCA scores. This potentially could be explained by the positioning of the voxel on the medial frontal cortex in the present study. Aβ_1–42 is associated to myelin pathology [47] and abnormal neuroinflammation [10], and affects frontal regions in later stages of development of AD.

Additionally, in group comparisons, our results show that MCI patients showed higher levels of frontal mI/tCr and Glx than BM- and BM+ subjects. Importantly, mI/tCr and Glx were not different between BM- and BM+, suggesting that higher expression of Aβ_1–42, p-tau, and t-tau may not influence these metabolites in the frontal cortex. It may also be that Aβ_1–42 is not concentrated in the frontal cortex until latter stages of AD pathology. However, these results are in line with previous findings showing increases in mI/tCr in the frontal cortex [48] and posterior cingulate [21, 37] of MCI patients compared to controls. This result also supports the idea of elevated neuroinflammation in the aging brain and in AD pathology, although the link between mI and inflammation has to be established more firmly [11]. Results in the literature regarding glutamatergic compounds in MCIs are more scattered: studies have shown decreased [45, 46] and increased [15] levels of glutamate and Glx in the posterior cingulate cortex or posterior regions. In the medial temporal lobe, Walecki and colleagues (2011) report an increase in Glx in MCI patients that progressed to AD. Excitotoxicity related to increased extracellular glutamate release has been depicted as parallel mechanism of age-related neuroinflammation. However, the relationship with MRS-measured Glx has to be defined more clearly and separation of signal for glutamate and glutamine is not reliable at 3T. Nevertheless, taken together higher levels of Glx and lower levels of GABA+ in MCI patients suggests an imbalance in excitatory and inhibitory function, a possible related factor to amyloid pathology and neuroinflammation in AD [49]. GABAergic dysfunction may begin in early stages of pathogenesis and could potentiate glutamatergic excitotoxicity [50]. Additionally, the fact that the sample in this present study is exclusively female cannot be overstated. Previously mentioned results from the literature were obtained in mixed samples, and sex was not unanimously used as covariate in analyses. Robust results from the literature suggest that neuroinflammation and microglial function differ with sex [51]. This could hypothetically alter MRS metabolites and should be investigated further with a larger sample size.

Several limitations in our work should be acknowledged, beginning with the uneven sample sizes in our groups. Given the trends observed in the MCI group, it is warranted to obtain a larger cohort in future comparative studies to strengthen the statistical power of comparisons between groups. Another limitation resides in not controlling for hormonal factors and blood-levels of estrogens. Given that our sample consists exclusively of older women, this should have been considered as an additional biological covariate in the study design, although the sample consisted of menopaused women. The effects of sex hormones in neurological development can generate 1) vulnerability to developing AD in females and 2) distinct sex differences in AD pathogenesis [43]. This highlights the importance of including sex and gender in aging research as a standard benchmark, but also in clinical decision-making and optimization of patient care [52]. A technical limitation of this study is that we did not control for T2 relaxation times of water and metabolites when comparing groups. As another limitation, we did not collect a separate spectrum of macromolecules and instead utilized the built-in macromolecular basis set from LCModel. Although the impact of age on macromolecule content is disputed [53, 54], a subject-specific macromolecule utilized during analysis would improve precision of quantification of all metabolites. Our results show a decrease of GABA+ with age, which is a replication of previous studies [13 , 55]. If macromolecular content increases with age and GABA+ contains some contribution of macromolecules, our aging-related decrease in GABA+ may be potentially underestimated, but it is unlikely that GABA+ is constant with aging, a notion supported by recent MRS meta-analytical evidence [56]. Future works combining other imaging modalities (e.g., cortical microstructure evidence [57]), with MRS with more robust clinical sample sizes are warranted to further the investigation of sex-based differences in MCI. Finally, in the present study, MoCA was the sole assessment obtained in both clinical and non-clinical (BM-, BM+) cohorts. It should be acknowledged that additional neuropsychological assessments should be used in future investigations, as the MoCA has limited sensitivity.

Conclusion

In this study, we have found that levels of frontal GABA+ significantly decrease with age and this relationship is significant independently of MCI pathology in women, and that levels of frontal tNAA/mI were correlated to general cognition only in women with MCI. To the best of our knowledge, the current study is the first to demonstrate a strong association between frontal GABA+ levels and neurological aging in a sample consisting exclusively of healthy older women and women with MCI. Importantly, our results show no correlation between CSF biomarkers and MRS metabolites in this sample.

Footnotes

ACKNOWLEDGMENTS

The authors wish to mention the support of the Goizueta Foundation in the Emory Brain Imaging Project and Emory Brain Health Study.

AHB is supported by an award from the Fonds de Recherche en Santé du Québec.

This work was supported by a gift from the Goizueta Foundation, NIH R01 AG070937 (JJL), and Roche IIS RD004723 (JJL). BC is supported by a Senior Research Career Scientist Award #B6364-L from the US Department of Veterans Rehabilitation Research & Development Service. This work was supported by National Institutes of Health grants (P30AG066511, R01AG070937, R01AG072603, and R21AG06440502).

Authors’ disclosures available online ().

References

(2021) 2021 Alzheimer’s disease facts and figures. Alzheimers Dement 17, 327–406.

Mielke

(2018) Sex and gender differences in Alzheimer’s disease dementia. Psychiatr Times 35, 14.

Andrew

, Tierney

(2018) The puzzle of sex, gender and Alzheimer’s disease: Why are women more often affected than men? Womens Health 14, 1745506518817995.

Altmann

, Tian

, Henderson

, Greicius

, Alzheimer’s Disease Neuroimaging Initiative Investigators (2014) Sex modifies the APOE-related risk of developing Alzheimer disease. Ann Neurol 75, 563–573.

Beydoun

, Boueiz

, Abougergi

, Kitner-Triolo

, Beydoun

, Resnick

, O’Brien

, Zonderman

(2012) Sex differences in the association of the apolipoprotein E epsilon 4 allele with incidence of dementia, cognitive impairment, and decline.720-731. e. Neurobiol Aging 33, 720–731.e724.

Nebel

, Aggarwal

, Barnes

, Gallagher

, Goldstein

, Kantarci

, Mallampalli

, Mormino

, Scott

, Yu

(2018) Understanding the impact of sex and gender in Alzheimer’s disease: A call to action. Alzheimers Dement 14, 1171–1183.

Oveisgharan

, Arvanitakis

, Yu

, Farfel

, Schneider

, Bennett

(2018) Sex differences in Alzheimer’s disease and common neuropathologies of aging. Acta Neuropathol 136, 887–900.

Ittner

, Götz

(2011) Amyloid-β and tau—a toxic pas de deux in Alzheimer’s disease. Nat Rev Neurosci 12, 67–72.

Kinney

, Bemiller

, Murtishaw

, Leisgang

, Salazar

, Lamb

(2018) Inflammation as a central mechanism in Alzheimer’s disease. Alzheimers Dement (N Y) 4, 575–590.

10.

Heneka

, Carson

, El Khoury

, Landreth

, Brosseron

, Feinstein

, Jacobs

, Wyss-Coray

, Vitorica

, Ransohoff

(2015) Neuroinflammation in Alzheimer’s disease. Lancet Neurol 14, 388–405.

11.

Rae

(2014) A guide to the metabolic pathways and function of metabolites observed in human brain 1 H magnetic resonance spectra. Neurochem Res 39, 1–36.

12.

Haga

, Khor

, Farrall

, Wardlaw

(2009) A systematic review of brain metabolite changes, measured with 1H magnetic resonance spectroscopy, in healthy aging. Neurobiol Aging 30, 353–363.

13.

Gao

, Edden

, Li

, Puts

, Wang

, Liu

, Zhao

, Wang

, Bai

, Zhao

(2013) Edited magnetic resonance spectroscopy detects an age-related decline in brain GABA levels. Neuroimage 78, 75–82.

14.

Nedelska

, Przybelski

, Lesnick

, Schwarz

, Lowe

, Machulda

, Kremers

, Mielke

, Roberts

, Boeve

(2017) 1H-MRS metabolites and rate of β-amyloid accumulation on serial PET in clinically normal adults. Neurology 89, 1391–1399.

15.

Fayed

, Modrego

, Rojas-Salinas

, Aguilar

(2011) Brain glutamate levels are decreased in Alzheimer’s disease: A magnetic resonance spectroscopy study. Am J Alzheimers Dis Other Demen 26, 450–456.

16.

Kantarci

, Weigand

, Przybelski

, Preboske

, Pankratz

, Vemuri

, Senjem

, Murphy

, Gunter

, Machulda

(2013) MRI and MRS predictors of mild cognitive impairment in a population-based sample. Neurology 81, 126–133.

17.

Pilatus

, Lais

, de Rochmont

AdM

, Kratzsch

, Frölich

, Maurer

, Zanella

, Lanfermann

, Pantel

(2009) Conversion to dementia in mild cognitive impairment is associated with decline of N-actylaspartate and creatine as revealed by magnetic resonance spectroscopy. Psychiatry Res 173, 1–7.

18.

Modrego

, Fayed

(2011) Longitudinal magnetic resonance spectroscopy as marker of cognitive deterioration in mild cognitive impairment. Am J Alzheimers Dis Other Demen 26, 631–636.

19.

Targosz-Gajniak

, Siuda

, Wicher

, Banasik

, Bujak

, Augusciak-Duma

, Opala

(2013) Magnetic resonance spectroscopy as a predictor of conversion of mild cognitive impairment to dementia. J Neurol Sci 335, 58–63.

20.

Porges

, Woods

, Edden

, Puts

, Harris

, Chen

, Garcia

, Seider

, Lamb

, Williamson

(2017) Frontal gamma-aminobutyric acid concentrations are associated with cognitive performance in older adults. Biol Psychiatry Cogn Neurosci Neuroimaging 2, 38–44.

21.

Kantarci

, Lowe

, Przybelski

, Senjem

, Weigand

, Ivnik

, Roberts

, Geda

, Boeve

, Knopman

(2011) Magnetic resonance spectroscopy, β-amyloid load, and cognition in a population-based sample of cognitively normal older adults. Neurology 77, 951–958.

22.

Wang

, Tan

, Wang

H-F

, Liu

, Yin

R-H

, Wang

W-Y

, Chang

X-L

, Jiang

, Yu

J-T

(2015) Magnetic resonance spectroscopy in Alzheimer’s disease: Systematic review and meta-analysis. J Alzheimers Dis 46, 1049–1070.

23.

Tumati

, Martens

, Aleman

(2013) Magnetic resonance spectroscopy in mild cognitive impairment: Systematic review and meta-analysis. Neurosci Biobehav Rev 37, 2571–2586.

24.

Spychala

, Honarpisheh

, McCullough

(2017) Sex differences in neuroinflammation and neuroprotection in ischemic stroke. J Neurosci Res 95, 462–471.

25.

Gaignard

, Liere

, Thérond

, Schumacher

, Slama

, Guennoun

(2017) Role of sex hormones on brain mitochondrial function, with special reference to aging and neurodegenerative diseases. Front Aging Neurosci 9, 406.

26.

McEwen

, Milner

(2017) Understanding the broad influence of sex hormones and sex differences in the brain. J Neurosci Res 95, 24–39.

27.

O’Gorman

, Michels

, Edden

, Murdoch

, Martin

(2011) In vivo detection of GABA and glutamate with MEGA-PRESS: Reproducibility and gender effects. J Magn Reson Imaging 33, 1262–1267.

28.

McKnight

, Wellons

, Sites

, Roth

, Szychowski

, Halanych

, Cushman

, Safford

(2011) Racial and regional differences in age at menopause in the United States: Findings from the REasons for Geographic And Racial Differences in Stroke (REGARDS) study. Am J Obstet Gynecol 205, 353.e1–353.e8.

29.

Goetz

, Hanfelt

, John

, Bergquist

, Loring

, Quyyumi

, Clifford

, Vaccarino

, Goldstein

, Johnson

2nd (2019) Rationale and design of the Emory Healthy Aging and Emory Healthy Brain studies. Neuroepidemiology 53, 187–200.

30.

Gruetter

(1993) Automatic, localized in vivo adjustment of all first-and second-order shim coils. Magn Reson Med 29, 804–811.

31.

Gruetter

, Tkáč

(2000) Field mapping without reference scan using asymmetric echo-planar techniques. Magn Reson Med 43, 319–323.

32.

Marshall

, Gatchel

, Donovan

, Muniz

, Schultz

, Becker

, Chhatwal

, Hanseeuw

, Papp

, Amariglio

(2019) Regional tau correlates of instrumental activities of daily living and apathy in mild cognitive impairment and Alzheimer’s disease dementia. J Alzheimers Dis 67, 757–768.

33.

Stoeckel

, Stewart

, Griffith

, Triebel

, Okonkwo

, Den Hollander

, Martin

, Belue

, Copeland

, Harrell

(2013) MRI volume of the medial frontal cortex predicts financial capacity in patients with mild Alzheimer’s disease. Brain Imaging Behav 7, 282–292.

34.

Provencher

(2001) Automatic quantitation of localized in vivo 1H spectra with LCModel. NMR Biomed 14, 260–264.

35.

Mullins

, McGonigle

, O’Gorman

, Puts

, Vidyasagar

, Evans

, Edden

(2014) Current practice in the use of MEGA-PRESS spectroscopy for the detection of GABA. Neuroimage 86, 43–52.

36.

Near

, Harris

, Juchem

, Kreis

, Marjańska

, Öz

, Slotboom

, Wilson

, Gasparovic

(2021) Preprocessing, analysis and quantification in single-voxel magnetic resonance spectroscopy: Experts’ consensus recommendations. NMR Biomed 34, e4257.

37.

Kantarci

, Smith

, Ivnik

, Petersen

, Boeve

, Knopman

, Tangalos

, Jack

Jr (2002) 1H magnetic resonance spectroscopy, cognitive function, and apolipoprotein E genotype in normal aging, mild cognitive impairment and Alzheimer’s disease. J Int Neuropsychol Soc 8, 934.

38.

Harris

, Puts

, Edden

(2015) Tissue correction for GABA-edited MRS: Considerations of voxel composition, tissue segmentation, and tissue relaxations. J Magn Reson Imaging 42, 1431–1440.

39.

Shaw

, Vanderstichele

, Knapik-Czajka

, Figurski

, Coart

, Blennow

, Soares

, Simon

, Lewczuk

, Dean

(2011) Qualification of the analytical and clinical performance of CSF biomarker analyses in ADNI. Acta Neuropathol 121, 597–609.

40.

Blennow

, Shaw

, Stomrud

, Mattsson

, Toledo

, Buck

, Wahl

, Eichenlaub

, Lifke

, Simon

(2019) Predicting clinical decline and conversion to Alzheimer’s disease or dementia using novel Elecsys Aβ (1–42), pTau and tTau CSF immunoassays. Sci Rep 9, 19024.

41.

Nasreddine

, Phillips

, Bédirian

, Charbonneau

, Whitehead

, Collin

, Cummings

, Chertkow

(2005) The Montreal Cognitive Assessment, MoCA: A brief screening tool for mild cognitive impairment. J Am Geriatr Soc 53, 695–699.

42.

Benjamini

, Hochberg

(1995) Controlling the false discovery rate: A practical and powerful approach to multiple testing. J R Stat Soc Series B Methodol 57, 289–300.

43.

Pike

(2017) Sex and the development of Alzheimer’s disease. J Neurosci Res 95, 671–680.

44.

Murari

, Liang

DR-S

, Ali

, Chan

, Mulder-Heijstra

, Verhoeff

NPL

, Herrmann

, Chen

, Mah

(2020) Prefrontal GABA levels correlate with memory in older adults at high risk for Alzheimer’s disease.tgaa. Cereb Cortex Commun 1, 022.

45.

Oeltzschner

, Wijtenburg

, Mikkelsen

, Edden

, Barker

, Joo

, Leoutsakos

J-MS

, Rowland

, Workman

, Smith

(2019) Neurometabolites and associations with cognitive deficits in mild cognitive impairment: A magnetic resonance spectroscopy study at 7 Tesla. Neurobiol Aging 73, 211–218.

46.

Riese

, Gietl

, Zölch

, Henning

, O’Gorman

, Kälin

, Leh

, Buck

, Warnock

, Edden

(2015) Posterior cingulate γ-aminobutyric acid and glutamate/glutamine are reduced in amnestic mild cognitive impairment and are unrelated to amyloid deposition and apolipoprotein E genotype. Neurobiol Aging 36, 53–59.

47.

Dean

, Hurley

, Kecskemeti

, O’Grady

, Canda

, Davenport-Sis

, Carlsson

, Zetterberg

, Blennow

, Asthana

(2017) Association of amyloid pathology with myelin alteration in preclinical Alzheimer disease. JAMA Neurol 74, 41–49.

48.

Walecki

, Barcikowska

, Ćwikła

, Gabryelewicz

(2011) N-acetylaspartate, choline, myoinositol, glutamine and glutamate (glx) concentration changes in proton MR spectroscopy (1H MRS) in patients with mild cognitive impairment (MCI).MT. Med Sci Monit 17, 105.

49.

, Wen

, Wu

, Shen

(2020) GABAergic dysfunction in excitatory and inhibitory (E/I) imbalance drives the pathogenesis of Alzheimer’s disease. Alzheimers Dement 16, 1312–1329.

50.

Calvo-Flores Guzmán

, Vinnakota

, Govindpani

, Waldvogel

, Faull

, Kwakowsky

(2018) The GABAergic system as a therapeutic target for Alzheimer’s disease. J Neurochem 146, 649–669.

51.

Hanamsagar

, Bilbo

(2016) Sex differences in neurodevelopmental and neurodegenerative disorders: Focus on microglial function and neuroinflammation during development. J Steroid Biochem Mol Biol 160, 127–133.

52.

Bartz

, Chitnis

, Kaiser

, Rich-Edwards

, Rexrode

, Pennell

, Goldstein

, O’Neal

, LeBoff

, Behn

(2020) Clinical advances in sex-and gender-informed medicine to improve the health of all: A review. JAMA Intern Med 180, 574–583.

53.

Marjańska

, Deelchand

, Hodges

, McCarten

, Hemmy

, Grant

, Terpstra

(2018) Altered macromolecular pattern and content in the aging human brain. NMR Biomed 31, e3865.

54.

Hui

, Gong

, Zöllner

, Song

, Murali-Manohar

, Oeltzschner

, Mikkelsen

, Tapper

, Chen

, Saleh

(2022) The macromolecular MR spectrum does not change with healthy aging. Magn Reson Med 87, 1711–1719.

55.

Hone-Blanchet

, Bohsali

, Krishnamurthy

, Shahid

, Lin

, Zhao

, Loring

, Goldstein

, John

, Fleischer

(2022) Relationships between frontal metabolites and Alzheimer’s disease biomarkers in cognitively normal older adults. Neurobiol Aging 109, 22–30.

56.

Porges

, Jensen

, Foster

, Puts

(2021) The trajectory of cortical GABA across the lifespan, an individual participant data meta-analysis of edited MRS studies. Elife 10, e62575.

57.

Hone-Blanchet

, Shahid

, Krishnamurthy

, Lin

, Levey

, Lah

, Qiu

, Crosson

(2020) CO-registration of proton magnetic resonance spectroscopy (1H-MRS) and diffusion imaging (DMRI) in healthy aging: A prospective study of cerebral neurochemistry and microstructure: Neuroimaging/Normal brain aging. Alzheimers Dement 16, e043251.