Early neurotransmission impairment in non-invasive Alzheimer Disease detection

Alzheimer Disease (AD) is a pathology suffered by millions of people worldwide and it has a great social and economic impact. Previous studies reported a relationship between alterations in different amino acids and derivatives involved in neurotransmission systems and cognitive impairment. Therefore, in this study the neurotransmission impairment associated to early AD has been evaluated. For this purpose, different amino acids and derivatives were determined in saliva samples from AD patients and healthy subjects, by means of an analytical method based on chromatography coupled to tandem mass spectrometry. Results showed statistically significant differences in salivary levels for the compounds myo-inositol, creatine and acetylcholine; and other compounds (myo-inositol, glutamine, creatine, acetylcholine) showed significant correlations with some cognitive tests scores. Therefore, these compounds were included in a multivariate analysis and the corresponding diagnosis model showed promising indices (AUC 0.806, sensitivity 61%, specificity 92%). In conclusion, some amino acids and derivatives involved in neurotransmission impairment could be potential biomarkers in early and non-invasive AD detection.


Scientific RepoRtS
| (2020) 10:16396 | https://doi.org/10.1038/s41598-020-73362-z www.nature.com/scientificreports/ Material and methods participants selection and samples collection. The study participants were patients from the Neurology Service in the University and Polytechnic Hospital La Fe (Valencia, Spain) who previously signed the informed consent. The experimental protocol for the study was approved by the Ethics Committee (CEIC) of the Health Research Institute La Fe (Valencia, Spain), and it is in accordance with the appropriate guidelines (Declaration of Helsinki). Participants were classified into AD (including mild cognitive impairment (MCI) due to AD (n = 17) and mild to moderate dementia due to AD (n = 14) and healthy control (HC, (n = 12)) groups. For this, they were subjected to neuropsychological tests (Repeatable Battery for the Assessment of Neuropsychological Status (RBANS), Clinical Dementia Rating (CDR), Minimental State Examination (MMSE), Functional Activities Questionnaire (FAQ)) [27][28][29][30] , structural neuroimaging by means of nuclear magnetic resonance (NMR) or computerized axial tomography (CAT) 31 , and cerebrospinal fluid (CSF) biomarkers (β-amyloid peptide (Aβ), total Tau (t-Tau), phosphorylated Tau (p-Tau)) 32,33 . Specifically, in the AD group, the MCI-AD participants showed cognitive complaints without daily living activities impairment, while mild dementia-AD participants showed minor daily living activities impairment. This classification was carried out following the National Institute on Aging-Alzheimer's Association (NIA-AA) recommendations [34][35][36] . Also, all AD participants showed positive levels of CSF biomarkers (Aβ < 700 pg mL -1 , t-Tau > 380 pg mL -1 , p-Tau > 70 pg mL -1 ) and altered neuropsychological evaluation (CDR > 0, RBANS.DM < 65, MMSE < 27). However, the HC group was characterized by negative CSF biomarkers, and normal neuropsychological evaluation. Regarding exclusion criteria, patients with a history of structural brain disease (tumour, stroke, etc.), major head trauma, epilepsy, multiple sclerosis and major psychiatric disorders were excluded, as well as patients not able to undergo neuropsychological evaluations.
Saliva samples were obtained from all the participants. They were whole-mouth saliva and collected by spitting into sterile bottles (without additives) between 10 and 12 a.m. (minimum 30 min after breakfast). Participants rinsed their mouth before saliva collection, between 1-2 mL could be collected across subjects with minor differences between control and case subjects. Then, the samples were aliquoted into 2 mL tubes, and those with visible blood contamination were excluded from the study. Finally, samples were stored at -80ºC until the analysis.
Sample treatment. The saliva sample (150 µL) was added to 300 µl of acetonitrile and centrifuged at 1200 g, 5 min at 4ºC. Then, 5µL of the internal standard (IS) solution (Phe-D5, 10 mmol L -1 ) were added to 95 µL of supernatant. The IS is an amino acid added as corrector of injection volume and ionization procedure. The range of calibration to quantify the analytes in samples was 0.3-5000 µmol L -1 . Finally, the samples were injected in the Ultra Performance Liquid Chromatography-tandem Mass Spectrometry (UPLC-MS/MS) system. Simultaneously, salivary total proteins were determined in each sample (5 µL) using a colorimetric protein assay kit (Pierce BCA) to standardize the analytes concentrations results. In general, these concentrations vary depending on saliva flow. So, the use of total proteins as corrective index approximates salivary flow rate adjustment 37,38 . UPLC-MS/MS analysis. The chromatographic system was a Waters Acquity UPLC-XevoTQ system (Milford, MA, USA) with triple quadrupole as mass analyzer. The instrumental conditions were positive electrospray ionization (ESI), capillary voltage 3.50 kV, extractor 5.00 V, source temperature 120 °C, desolvation temperature 350 °C, nitrogen cone flow 50 L/h and desolvation gas flow 750 L/h. Separation conditions were selected to achieve appropriate chromatographic retention and resolution by using an HILIC column (100 × 2.1 mm, 7 µm, 100 Å) from Phenomenex. Mobile phase used was CH3OH (5 mmol L -1 NH4HCO2): H2O (5 mmol L -1 NH4HCO2), (70:30) with isocratic gradient 10 min. The flow rate, column temperature and injection volume were set at 0.4 mL/min, 30°, and 5 µL, respectively.
During batch analysis, samples were kept at 4 °C. Mass spectrometric detection was carried out by multiple reaction monitoring (MRM) (see Table 1) (all 10 compounds analysed in a single MRM run).
The data were acquired and processed using the MassLynx 4.1 and QuanLynx 4.1 (Waters) software, respectively. Peak area integration was used, and analytes responses were expressed as ratios to the Phe-D5 (IS) in all standards and samples.

Method validation.
The validation procedure consisted of the assessment of some analytical characteristics (linearity, precision, accuracy, limit of detection (LOD), limit of quantification (LOQ), stability). The individual standards were prepared in H 2 O, and the calibration standards in CH 3 CN:H 2 O (50:50, v/v). Standards were run in triplicate. The linearity was evaluated constructing a calibration curve (0.1-5000 nmol L -1 ) for each analyte (n = 9). The precision was estimated from standards at mid concentration level (1000 nmol L -1 for each analyte) within one validation batch (intra-day) and among validation batches (inter-day, along 1 month) (by triplicate). The accuracy was evaluated by means of the recovery test. For that, saliva samples were spiked at three concentration levels (low, mid, high), and they were analyzed each of the three validation days. The LODs and LOQs were established, as the concentrations generating a signal-to-noise ratio of 3and 10, respectively. Analytes stability after three freeze-thaw cycles was assessed by means of a spiked saliva sample (1000 nmol L-1, each analyte). Chicago, IL, USA), and for the multivariant analysis Unscrambler software v7.6 (Norway) was used. Differences between AD and healthy control participants were evaluated by means of t-test for numerical variables with normal distribution (Kolmogorov-Smirnov test), expressing results as mean ± standard deviation. While, Mann-Whitney and Chi-square non-parametric tests were used for numerical and categorical variables without normal distribution, expressing results as median and interquartile range (IQR). In addition, correlations between salivary analytes and standard neuropsychological scales, as well as CSF biomarkers were analysed by Pearson Correlation. In all the cases, statistical significance was fixed in a p value of 0.05. Multivariate analysis was carried out by means of Partial Least Square (PLS) using some dependent variables (metabolites) and one independent variable (participant's group). Then, the Receiver operating characteristic curve (ROC) of the model was obtained.

Results
Analytical performance data. The analytical method was validated by employing standards containing all the analytes in the 0.1-5000 nmol L -1 concentrations range. Results are summarized in Supplementary Material (Table S1). The method provided an adequate linearity for all the analytes (R 2 between 0.996 and 0.999), and satisfactory sensitivity (LODs between 0.09 and 1.4 nmol L -1 ). Also, suitable precision was obtained with intra-day and inter-day coefficients of variation of 0.8-3.2% (n = 3) and 2.9-10.2% (n = 6), respectively (at a concentration of 1000 nmol L -1 ). The accuracy of the method was evaluated by analysing spiked saliva samples containing analytes at different concentrations (low, medium and high) within the tested concentration ranges. The results obtained are listed at Table S2 (Supplementary Material), quantitative recoveries were achieved for all the analytes. The analytes stability was assayed after three freeze thaw cycles, determining the concentrations in spiked saliva samples (1000 nmol L -1 , each analyte) by triplicate. The recoveries were between 88 and 106%. So, no significant deterioration of the analytes was observed.
Demographic and clinic participants description. Participants' demographic and clinical data are shown in Table 2. As it is expected, statistically significant differences were observed between participants groups for neuropsychological variables (RBANS, CDR, FAQ, MMSE) and CSF biomarkers (β-amyloid, Tau, p-Tau).
Amino acids and derivatives impairment in Alzheimer Disease. Different amino acids and derivatives (aspartic acid, glutamic acid, glutamine, GABA, creatine, taurine, N-acetil aspartate, myo-inositol, acetylcholine) were determined in saliva samples from AD patients at early stages of the pathology (MCI and mild dementia), and from HC participants ( Table 3). As result, salivary myo-inositol (p = 0.018) and creatine (p = 0.049) showed lower levels in AD compared to control group, while acetylcholine showed higher levels in AD patients (p = 0.015) (Fig. 1). Not statistically significant differences were found for the other compounds. Also, the compounds levels were analysed in function of gender, observing that only glutamine showed statistically significant differences (p = 0.031).
Then, different diagnosis models were developed to compare individual accuracy of the selected compounds, as they showed statistically significant differences between AD and HC groups, as well as correlations with some relevant clinical variables used in AD diagnosis.  Multivariate analysis.. Different combinations of the previously selected compounds (myo-inositol, creatine, glutamine, acetylcholine) were used to carry out multivariate analysis. In general, the developed models showed AUC-ROC values between 0.382 and 0.806 (see Table 4). However, model 1 (myo-inositol, creatine, glutamine, acetylcholine) showed best performance. The results from this multivariate analysis model are depicted in Fig. 2. As can be seen the salivary levels of myo-inositol, creatine, glutamine and acetylcholine could dis-    Figure 2. PLS score plot represents differential distribution between AD and healthy control groups. As visual approach, blue circle englobes AD patients, green circle englobes healthy and some AD participants. PC: principal component.
Regarding the predictive power of this model it is important to highlight the specificity of 92%, while the sensitivity was 61%.

Discussion
Saliva sample has the advantage of simple collection, being convenient and acceptable for all patients, and achieving high participant recruitment. In addition, several studies have found that saliva levels were comparable to blood levels, adding some clinical value to these salivary determinations 39 . The salivary levels of neurotransmitters were evaluated from two different approaches. First, univariate analysis showed significant differences between AD and HC participants for some compounds (myo-inositol, creatine, acetylcholine), as well as, correlations with clinical AD variables (neuropsychological scales, CSF biomarkers) for other compounds (myo-inositol, glutamine, creatine). Second, multivariate analysis, including the previously selected compounds (myo-inositol, glutamine, creatine, acetylcholine), showed satisfactory accuracy discriminating between AD and healthy control individuals. Regarding univariate analysis, glutamine showed negative correlation with MMSE score. Similarly, previous studies showed increased CSF levels of glutamate and glutamine in patients with probable AD 24 . In addition, the glutamate/glutamine ratio showed a reduction in ageing and AD 40 . It could be explained by the glutamine neuroprotector effect observed in cell culture, protecting against amyloid-β peptide 41 . Also, in the present study myo-inositol levels were lower in AD patients. Specifically, salivary myo-inositol showed lower levels in saliva samples from patients suffering from important cognitive impairment, while higher salivary levels corresponded to higher neuropsychological tests scores (RBANS, MMSE). In this sense, Shinno et al. found lower levels of myoinositol in brain, specifically in the anterior cingulate gyros, and it could be associated with AD development 42 . In contrast, a previous study found that urinary myo-inositol excretion could be used as cognitive impairment biomarker 43 . In definitive, myo-inositol levels could be impaired under AD conditions, making difficult the transport of substances between brain and other biofluids 44 . In addition, salivary myo-inositol showed correlation with CSF β-amyloid levels, considered the gold standard in AD diagnosis. Similarly, previous studies showed that brain myo-inositol measured by magnetic resonance spectroscopy correlated with β-amyloid even before the cognitive impairment appearance 45 . In general, these findings could constitute a relevant aspect in non-invasive diagnosis development, since these amino acids and derivates measured in non-invasive samples (saliva, urine) could replace CSF sampling in AD current diagnosis. Moreover, other inositol stereoisomers, such as scylloinositol, had been tested as potential therapy because of its anti-oligomer activity in cell culture 46 . In this sense, an increase in myo-inositol levels could be protective against amyloid-β plaques formation and so AD development.
For acetylcholine, the present study showed higher levels in patients with lower scores for RBANS and higher scores for CDR and FAQ. Of note, CDR and FAQ lower scores show better cognitive status, while higher punctuations in RBANS show better cognitive state. Therefore, high salivary acetylcholine levels corresponded to patients with cognitive impairment. Nevertheless, opposite results were obtained in a previous study carried out in plasma samples 47 .
For creatine, lower salivary levels were obtained in AD patients, as well as positive correlation with MMSE. In this sense, a few studies in literature have focused on this metabolite as neurodegeneration biomarker 48,49 .
Regarding multivariate analysis, the studied neurotransmitters could be useful as AD biomarkers. Specifically, the simultaneous determination of a panel of 4 neurotransmitters (myo-inositol, acetylcholine, creatine, glutamine) could improve the accuracy differentiating between AD and healthy participants. In general, these results would explain an alteration in neurotransmission under AD conditions. A previous study in AD showed an imbalance in neurotransmitters levels in brain 50 . In addition, neurotransmitters' transporters showed alterations in the pathology 51 . Therefore, neurotransmitters were accumulated or reduced in brain and different biofluids depending on their transport, and an alteration in their distribution could lead to the typical manifestations of AD. In this sense, a previous AD study determining glutamate in CSF samples showed sensitivity 95.2%, specificity 100%, and AUC = 0.99 24 . Also, a study determining 10 amino acids as potential blood biomarkers in AD showed an AUC = 0.948 52 . Despite obtaining satisfactory diagnosis indexes, these works used invasive samples. Nevertheless, the present study could allow a first screening approach in early AD diagnosis, using non-invasive samples. In fact, this new tool, showing high positive predictive value, could reduce invasive and expensive diagnostic methods (CSF biomarkers, neuroimaging techniques) only to doubtful cases, avoiding conventional diagnosis methods in positive screening cases.
Regarding specificity, some studies found neurotransmitter impairment in other dementias (Parkinson disease, vascular dementia, dementia with Lewy bodies, fronto-temporal dementia) 53,54 , or pathologies 55 . Nevertheless, there is a lack of studies focusing on differential AD diagnosis based on neurotransmitters.
Among the study limitations, it is important to highlight the small sample size and so the reduced statistical power. Also, a single internal standard has been used to correct 10 different analytes which could impact in quantitative accuracy. In addition, some saliva collection variables have not been standardised, and potential biomarkers specificity has not been evaluated.
conclusions Some neurotransmitters levels in saliva could be impaired under AD conditions. Among the evaluated compounds, it is important to highlight myo-inositol, acetylcholine, creatine and glutamine, since these compounds correlated with cognitive impairment associated to AD, and also myo-inositol showed correlation with CSF amyloid-β levels. These neurotransmitters could be used as promising non-invasive biomarkers for AD. From them, an optimum multivariate model was developed constituting a potential tool in the early screening of AD Scientific RepoRtS | (2020) 10:16396 | https://doi.org/10.1038/s41598-020-73362-z www.nature.com/scientificreports/ Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.