Potential involvement of monoamine oxidase activity in SARS-COV2 infection and delirium onset
=============================================================================================

* Miroslava Cuperlovic-Culf
* Emma L. Cunningham
* Anu Surendra
* Xiaobei Pan
* Steffany Bennett
* Mijin Jung
* Bernadette McGuiness
* Anthony Peter Passmore
* Danny McAuley
* David Beverland
* Brian D. Green

## Abstract

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) causes a range of extra-respiratory signs and symptoms. One such manifestation is delirium, an acute confusional state occurring in 60-70% of severe SARS-CoV-2 cases. Delirium is also a common clinical syndrome following planned orthopedic surgery. This investigation initially explored the underlying role of metabolism in delirium-susceptibility in this setting. Metabolomics profiles of cerebrospinal fluid (CSF) and blood taken prior to surgery found significant concentration differences of several amino acids, acylcarnitines and polyamines were observed in delirium-prone patients. Phenethylamine (PEA) concentrations in delirium-prone patients was significantly lower in CSF than in blood, whilst in age- and gender-matched controls the opposite was observed (adjusted p values: 1.8×10−6 (control) and 1.788×10−10 (delirium)). PEA is metabolised by monoamine oxidase B (MAOB), a putative enzyme target for the treatment of Alzheimer’s disease, Parkinson’s disease and depression. Our computational structural comparisons of MAOB and angiotensin converting enzyme (ACE) 2 found high similarity, specifically within the SARS-CoV-2 spike protein. MAOB structural alignment to ACE2 was 51% overall, but this was over 95% in the ACE2-spike protein binding region. Thus, it is possible that the spike protein interacts with MAOB on a molecular level. A previously published metabolomic dataset of control subjects and patients with either mild or severe COVID-19 was then analysed. Major concentration differences in some metabolites attributed to altered MAO activity were detected. Therefore, our hypothesis is that the SARS-CoV-2 influences MAOB activity, which is one potential cause for the many observed neurological and platelet based complications of SARS-CoV-2 infection.

**Significance Statement** Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) causes a range of extra-respiratory signs and symptoms including delirium, anosmia, intravascular coagulation. Analysis of metabolomics differences in delirium-prone patients and in severe cases of SARS-CoV2 infection indicate possibly significant role for monoamine oxidase B (MAOB). Changes in activity of MAOB have been shown in other conditions to cause range of symptoms observed in SARS-CoV2 infection. Computational structural analysis presented here demonstrations significant similarity between MAOB and ACE2 protein binding region leading to a hypothesis that SARS-CoV-2 spike protein affects MAOB activity during the infection. This proposition is of possibly highly significance for both diagnosis of high-risk subjects prior to infection as well as novel avenues for the development of treatment and prevention.

Keywords
*   SARS-COV2
*   Delirium
*   Monoamine oxidase
*   Metabolomics
*   Molecular modeling

## Main Text

## Introduction

COVID-19 is an ongoing major global health emergency caused by severe acute respiratory syndrome coronavirus SARS-CoV-2. Patients admitted to hospital with COVID-19 show a range of features including fever, anosmia, acute respiratory failure, kidney failure and gastrointestinal issues and the death rate of infected patients it is estimated at 2.2% (1). Some early studies of hospitalized patients indicated that 20-30% of COVID-19 patients develop some form of delirium or mental status change rising to 60-70% for those patients with severe illness of all age groups (2,3). The exact mechanisms are not understood although a number of possible causes have been proposed including direct viral invasion, cerebrovascular involvement, hypoxia, pyrexia, dehydration, hyperinflammation (cytokine storm), medications or metabolic derangements (2,3). The X-ray structure of SARS-CoV-2 spike protein binding to ACE2 (4,5) as well as recently demonstrated specific structural features of SARS-CoV-2 spike protein, (6) suggest specific features of SARS-CoV-2 spike protein structure for binding to human protein in the SARS-CoV-2 influence it’s virulence.

Delirium, an acute disorder of attention and cognition (American Psychiatric Association, 2013) is an unpleasant experience for patients, relatives and healthcare staff and is associated with negative outcomes such as dementia and death. As has been described by Fong et al. (7), whilst even those with the most resilient of brains can develop delirium in the face of severe stressors, delirium in the face of relatively minor insults may be a sign of underlying neurodegeneration (8). An improved understanding of the causes of postoperative delirium could provide a better appreciation of the understanding of the vulnerabilities causing this state following surgery, as well as following SARS-CoV-2 infection.

Planned orthopedic surgery under spinal anaesthesia provides a unique opportunity to preoperatively sample cerebrospinal fluid (CSF) samples in a group of patients where an estimated 17% will subsequently develop delirium (9). Metabolomics analysis of body fluids provides a wide-ranging molecular window into the major metabolic processes of the body. For the investigation of delirium a detailed examination of CSF provides an insight into altered brain metabolism. CSF metabolite biomarkers of delirium risk have already been identified (10). Major differences in the concentrations of polyamines (including spermidine and putrescine) are present in delirium-prone patients even before surgery or the presentation of delirium. However, it is not clear whether such metabolic changes occur more peripherally in the blood circulation, or whether the transfer of these metabolite across the blood-brain barrier is important. This investigation conducted an identical metabolomics methodology on the corresponding paired pre-operative blood samples comparing the individual blood and CSF profiles of each patient.

Recently, Shen et al. (10) presented metabolomics and proteomics analysis of serum of mild and severe COVID-19 patients indicating major concentration differences in serotonin, kynurenine, a number of amino acids, as well as, alterations in tryptophan and polyamine metabolic pathways. This metabolomics and proteomics analysis also indicated major role for platelets in SARS-CoV-2 infection. The metabolomics data were in agreement with clinical studies showing coagulopathy as one of the major issues in severe COVID-19 cases (12) as well as delirium, possibly associated with changes in neurotransmitters, such as serotonin (13). In the present investigation we interrogate different clinical metabolomics datasets (including one produced by our group) in order to uncover a biochemical relationship between delirium and COVID-19. The involvement of the enzyme monoamine oxidase (MAO) emerges as a common potential candidate. MAO is worthy of further exploration given that it is important for neurotransmitter metabolism, has prior associations with delirium, and is involved in platelet regulation and coagulation (14-16) as well as anosmia (17). MAO including MAOA and MAOB are flavoenzymes that catalyze the oxidative transformation of monoamines. The inhibition of these enzymes is a target for therapeutic development with the selectivity of candidate molecules for one isoform over another being a key consideration (20, 21). Alterations in the activities of MAOs is a potential source of various neuropsychiatric disorders including depression, autism or aggressive behavior (22, 23). Additionally function of MAOs represents an inherent source of oxidative stress, leading to the damage and death of neurons, ultimately leading to neurodegenerative diseases such as Parkinson’s or Alzheimer’s disease (24, 25).

## Results

### Delirium-associated alterations of blood and CSF metabolites

Principal component analysis (PCA) and 2-class Partial Least Squares-Discriminant Analysis (PLS-DA) were performed on CSF and blood metabolomic data samples. For both sample types PCA achieved only limited separation of control and delirium-prone cases (Supplementary Figure 1A and 1B). Incorporation of patient gender had very minor improvement of PCA group separation (Supplementary Figure 1C). Supervised PLS-DA identified combination of metabolite features capable of clearly distinguishing the two patient groups (Supplementary Figure 1C and 1D). For the CSF data the output was closely aligned with the previously published analysis (10). The importance of specific metabolites to the model’s discriminatory power are shown in coefficient plots (Supplementary Figure 2). The most important contributors in control and delirium sample separation in CSF were spermidine, putrescine, glutamine (as previously reported by Pan et al (10)), but also 3-hydroxypalmitoleylcarnitine (C16-1-OH), linoleic acid (C18-2), carnitine (C0) and, to a lesser extent, a number of other metabolites (Supplementary Figure 2 shown in red). For blood metabolomic data the discrimination of control and delirium patients was influenced most by levels of proline, ornithine, lysine, trans-4-hydroxyproline (t4-OH-Pro), PC aa C24:0 and PC aa C26:0 as well as H1, ADMA, C7-DC and, to a lesser extent, a number of other metabolites (shown in red in Supplementary Figure 2).

The above results were further corroborated by application of Statistical Analysis for Microarrays (SAM) (18), which has been previously applied to metabolomics data (19) (Figure 1A). SAM analysis of normalized CSF data found significant differences in CSF for ornithine, glutamine, putrescine, spermidine and threonine, similarly as previously shown (10) and in good agreement with the above PLS-DA (ornithine, glutamine, putrescine and spermidine were major features). Threonine was not one of the most significant contributors to the classification coefficients, but was still a significant contributor to the PLS-DA sample grouping. SAM analysis of blood metabolites shows major differences in pre-operative blood samples in proline, threonine, lysine, ornithine and phosphatidylcholines (specifically PC aa C26:0).

![Figure 1.](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2020/06/19/2020.06.16.20128660/F1.medium.gif)

[Figure 1.](http://medrxiv.org/content/early/2020/06/19/2020.06.16.20128660/F1)

Figure 1. 
A. SAM selection of the most relevant metabolites for classification control and delirium group from metabolites in pre-operative CSF and blood samples; B SAM analysis of major differences between CSF and blood metabolites in the control and delirium groups. In both cases analysis was performed following metabolite and sample normalization in TMeV. For PEA – Control, adjusted p-value 1.8E-6; for PEA – Delirium, adjusted p-value 1.78 E-10.

The availability of metabolic profiles for CSF and blood provides a unique possibility for the determination of significant differences in metabolite concentrations between these two body fluids giving information about the transfer and metabolism across the blood brain barrier as well as the possibility for the determination of blood-based biomarkers that are representative of changes in the CSF (Figure 1B). As can be expected there are major metabolic differences between CSF and blood in both control and delirium-prone group with majority of metabolites showing highly comparable behaviour in the two groups. Notable exceptions are metabolites that show opposite concentration in the two body fluids namely – phenethylamine (PEA) with higher concentration in CSF relative to blood in control subjects and opposite in delirium and Octadecadienylcarnitine (C 18:2) and hexanoylcarnitine (C6 (C4:1-DC)) showing higher concentration in blood then in CSF in control group. PEA is a natural monoamine alkaloid that acts as a central nervous system stimulant. Octadecadienylcarnitine (also called Linoleyl carnitine) and hexanoylcarnitine is a long-chain acyl fatty acid derivatives of carnitine. Difference in concentration in these three metabolites are significant in both control and delirium groups with adjusted p-values for blood to CSF groups in control: for PEA adj. p= 1.83e-6; C 18:2 adj. p= 6.1e-4 and C6 (C4:1-DC) adj. p= 6.6e-4 and delirium: PEA adj. p= 1.8e-10; C 18:2 adj. p= 2.8e-10 and C6 (C4:1-DC) adj. p= 5.9e-8. Although the difference in these three metabolites between CSF and blood is significant the difference of their concentrations between control and delirium observed separately in CSF and blood is only minor. In CSF PEA is overall slightly reduced in the delirium group (adj. p-value = 0.1) while C 18:2 and C6 (C4:1-DC) are unchanged. Similarly in plasma PEA concentration is slightly lower in delirium and C 18:2 and C6 (C4:1-DC) are unchanged. Therefore, overall PEA is slightly reduced in delirium-prone patients, prior to operation and significantly, in this group PEA concentration is significantly lower in CSF then in blood. PEA is a neurotransmitter integral to the signalling process in the central nervous system. It can be oxidised through the function of monoamine oxidase (MAO) enzymes.

Observed changes in PEA levels suggest changes in activity of MAO in delirium-prone patient cohort. MAOB is the only MAO protein found in platelets (for MAOA and MAOB protein expression please see Supplementary Figure 3) and has been linked to the activity of platelets and dysfunction of nitric oxide synthase pathway observed in number of neurological diseases (recently reviewed in (15))

In order to determine relationship between MAO and the related metabolites to the severity of SARS-CoV-2 response we have explored the dataset provided by (11) initially analysing the major metabolic differences between mild and severe COVID-19 cases. Patient information provided with the original publication does not include any data on possible delirium in these patients and we assumed, based on recent clinical studies (47) that patients with severe disease were more likely to have delirium than mild cases (with 60-70% in severe cases and 20-30% of hospitalized patients). For comparison we are providing major features selected using statistical, SAM methods (Figure 3A) as well as machine learning – Random Forest methodology and SHAP analysis of feature (metabolite) contribution to the ML model (Figure 3B).

![Figure 2.](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2020/06/19/2020.06.16.20128660/F2.medium.gif)

[Figure 2.](http://medrxiv.org/content/early/2020/06/19/2020.06.16.20128660/F2)

Figure 2. 
Metabolic Atlas analysis of interaction partners for PEA and putrescine. Many metabolites determined to be significantly altered in delirium-prone patients are closely linked to these two metabolites based on the metabolic network (46).

![Figure 3.](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2020/06/19/2020.06.16.20128660/F3.medium.gif)

[Figure 3.](http://medrxiv.org/content/early/2020/06/19/2020.06.16.20128660/F3)

Figure 3. 
Analysis of major metabolic differences between Mild and Severe COVID-19 cases based on (11) dataset. A. SAM analysis of major features following sample and metabolite normalization; B. SHAP analysis of major contributors to Random Forest classification of mild v.s. severe cases.

The number of metabolites showing major concentration difference between severe and mild cases can be related to pathways involving MAO enzyme (see Supplementary Figure 4 for all known direct interactions of MAOA and/or MAOB and metabolites). Specifically ratios of ceramide to sphingosine 1-phosphate (known as “sphingolipid rheostat”) is known to affect activity of MAO (26). The ratio between these metabolites based on the COVID-19 patient data provided by (11) is shown in Figure 4A and shows an increase with disease severity suggesting increasing activity of MAO. However, analysis of specific reaction partners that are related to MAO function shows a slight decrease of MAOB activity in severe cases (Figure 4B) while MAOA activity (indicated with Figure 4C reaction) is increasing in severe patient response.

![Figure 4.](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2020/06/19/2020.06.16.20128660/F4.medium.gif)

[Figure 4.](http://medrxiv.org/content/early/2020/06/19/2020.06.16.20128660/F4)

Figure 4. 
Median value and standard deviation of ratios between metabolites known to either activate MAO or to be part of metabolic activity of MAOA or B for healthy (H), mild (R) and severe (S) COVID-19 cases. A. Ratios of * ceramide (d18:1/20:0, d16:1/22:0, d20:1/18:0) / sphingosine 1-phosphate and ** ceramide (d18:2/24:1, d18:1/24:2) / sphingosine 1-phosphate. B. Part of Phenylalanine pathway with activity of MAOB showing ratio of 1. phenylalanine/phenylacetate; 2. phenylalanine/N-acetylphenylalanine and 3. phenylalanine/phenylpyruvate; C. part of tryptophan pathway where MAOA is used for serotonin metabolism. Shown is the median ratio of concentrations for serotonin/5-Hydroxy-indolacetate.

In order to investigate the possibility that the SARS-CoV-2 virus directly influences MAOB function the structures of MAOB and ACE2 (as a known binding target of SARS-CoV-2) were compared (4). Overall structural comparison between MAOB (PDB structure 1GOS) and ACE2 (PDB structure 6M0J), showed only 51% overall structural similarity. However the comparison of the ACE2 – spike protein binding region with MAOB resulted in 90% to 100% structure overlap. Further computational analysis of the spike-protein relation to MAOB protein is shown in Figure 5.

![Figure 5.](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2020/06/19/2020.06.16.20128660/F5.medium.gif)

[Figure 5.](http://medrxiv.org/content/early/2020/06/19/2020.06.16.20128660/F5)

Figure 5. 
A. analysis of the overall structure similarity between ACE2 and MAOB using jFATCAT_flexible (27) resulting in 6M0J Similarity:42% 1GOS Similarity:51% P-value:6.67e-01. Further analysis of structure subsection structural alignment in the regions of ACE2 related to binding Spike protein are shown in 1 and 2 showing major similarity (Section 1 6M0J Similarity:100% 1GOS Similarity:100% P-value:1.67e-02 Section 2 6M0J Similarity:74%; 1GOS Similarity:100% P-value:5.01e-01); B. Hypothetical binding between Spike protein and MAOB outlining the ligand location in the MAOB protein.

Specifically, analysis of the overall structure similarity between ACE2 and MAOB using jFATCAT_flexible (27) resulted in ACE2 similarity of 42% and MAOB similarity of 51% with the P-value:6.67e-01. However, further analysis of subsection of ACE2 involved in binding to Spike protein indicated as 1 and 2 (Figure 5A) shows major similarity (Section 1 6M0J Similarity:100% 1GOS Similarity:100% P-value:1.67e-02 Section 2 6M0J Similarity:74%; 1GOS Similarity:100% P-value:5.01e-01) suggesting possibility for interaction between MAOB and SARS-CoV-2 spike protein (Figure 5A shows the association between MAOB and ACE2 overlapping regions and spike protein) leading to possibility for MAOB activity change in severe COVID-19 patients (Figure 5B shows computational representation of the binding location of spike protein to MAOB indicating also the MAOB ligand).

## Discussion

Delirium is an acute confusional state occurring as a result of a precipitant such as certain medications, substance abuse, illness or surgery. Even routine surgical procedures, such as arthroplasty, are known to have post-operative delirium rates with 14% to 24% of elderly patients experiencing delirium after a routine surgery (28). A growing body of evidence links SARS-CoV-2 infection to delirium incidence (13). Here we explored whether there are common metabolic pathways linking these different triggers. This investigation generated and interrogated metabolomics data in order to identify the underlying metabolic perturbations associated with post-operative delirium. A range of metabolite associations emerged (Figure 1), but PEA was one of the most profoundly affected metabolites in patients who later experienced postoperative delirium. In delirium-prone individuals PEA concentrations are slightly reduced both in the CSF and blood plasma relative to the control group. Interestingly, for PEA an inverse relationship exists in these two compartments for these two groups with the significantly smaller PEA concentration in CSF relative to blood in delirium-prone cohort. Brain monoamines include common biogenic amines (dopamine, norepinephrine, and serotonin) and trace amines such as PEA (29). PEA has been shown to alter the serotonin transporter by interacting with trace amine-associated receptor 1 (TAAR1) (29). Activation of TAAR1 with PEA significantly inhibits uptake and induces efflux of dopamine and norepinephrine.

Observed alterations in PEA are highly plausible because monoamine oxidase (MAO), one of the key enzymes responsible for its metabolism, is a known target for treatment of a variety of neurological conditions including depression, Parkinson’s disease and recently Alzheimer’s disease (16). MAO is an enzyme localised on the outer mitochondrial membrane and it preferentially degrades benzylamine and PEA. The MAO family of proteins oxidizes a number of different amine substrates including small-molecule monoamines, polyamines as well as modified amino acids in proteins, and directly influences number of different metabolites (Supplementary Figure 4).

Two MAO subtypes exist: monoamine oxidase A (MAOA) which preferentially oxidizes biogenic amines such as 5-hydroxytryptamine (5-HT), norepinephrine and epinephrine, and monoamine oxidase B (MAOB) which performs oxidative deamination of biogenic and xenobiotic amines. MAOB is particularly important for the metabolism of neuroactive and vasoactive amines in the central nervous system and peripheral tissues. Expression of MAOB increases with age and is associated with increases in free radical damage and ROS formation. This in turn leads to a decrease in neuronal mitochondrial function and ultimately neurodegeneration (30) which is partly due to reduced PEA concentrations.

MAO inhibitors have been extensively developed and utilized for treatment of depression (9, [https://www.nice.org.uk/](https://www.nice.org.uk/)). A number of publications have also investigated the therapeutic effects of MAO inhibitors for other neurological conditions, such as Alzheimer’s disease, Parkinson’s disease, and depression (16). Specifically, MAOB has been proposed as a possible therapeutic target for Alzheimer’s disease due to its association with aberrant GABA production, but it also has therapeutic relevance for Parkinson’s disease due to its role in dopamine depletion (16, 31).

Utilization of MAO inhibitors (MAOI) for depression treatment has resulted in a number of side effects including: agitation, irritability, ataxia, movement disorders, insomnia, drowsiness, vivid dreams, cognitive impairment, and slowed speech, hallucinations and paranoid delusions ([https://www.nice.org.uk/](https://www.nice.org.uk/)). Additionally, MAOI has been linked to an increasing suicide, pyrexia, delirium, hypoxia, hypertension and fatal intravascular coagulation (32, 33, 34). MAOB is highly expressed in neurones, as well as platelets (35) possibly explaining the observed effects of MAOB inhibitors on neurological state as well as blood based complications (Supplementary Figure 3 shows protein expression of MAOA and MAOB in different human tissues).

Previously reported links between HIV infection and changes in monoamine and acylcarnitine metabolites (as well as inflammatory markers) indicates that viral infection and inflammation can alter monoamine metabolism and mitochondrial energetics (36). At the same time a number of side-effects previously listed for MAO inhibitors have also been observed in COVID-19 patients. One of the, as yet unresolved, effects of SARS-CoV-2 infection in a subgroup of patients is the development of a systemic coagulopathy and acquired thrombophilia characterized by a proclivity for venous, arterial and microvascular thrombosis (12). Additionally, severe cases of SARS-CoV-2 infection (where delirium is apparent) also experience low blood oxygen levels, elevated urea, acute renal dysfunction (37) – all of which are symptomatic of MAOB inhibition overdose or drug side-effects such as anosmia, a known symptom of dopamine depletion in Parkinson’s disease. Differences in the activity of MAOB in surgical delirium-prone patient population was indicated with change in concentration of PEA as well as polyamines. Significant concentration increases in severely affected COVID-19 patients is observed for both indolacetate and kynurenine as well as apparent increase in metabolising of serotonin in severe cases (Figure 4) all part of tryptophan metabolism utilizing MAOA. Additionally, tryptophan plasma concentration in both mild and severe patients is significantly reduced relative to the healthy subjects (Supplementary Figure 5) corroborated with the reduction in the concentration of arachidonate and related metabolites in severe relative to mild and both groups of infected patients relative to healthy subjects. Arachidonic acid is metabolized by activated platelets possibly leading to their aggregation. High risk groups for severe response to SARS-CoV-2 infection have known increased activity of MAO enzymes including: age, obesity, diabetes, heart condition ([www.who.int](http://www.who.int)). In addition to other symptoms, SARS-CoV-2 causes hematological changes which include reduced platelet count (38). With known role of MAO in number of observed SARS-CoV-2 symptoms we have performed computational analysis of possible effect of SARS-Co2 virus spike protein on MAO. A preliminary computational structure comparison of MAOB and ACE2 protein has been performed here in order to determine whether there is a possibility for binding of the SARS-CoV-2 spike protein to MAOB (Figure 5 and Supplementary Figure 6-9). Although the sequence similarity between ACE2 and MAOB proteins is limited there is an almost coincident alignment and structure similarity in the region involved in SARS-CoV-2 spike protein binding. Binding of the spike protein to MAOB could result in a change in either its enzymatic function, its post-translational modification or association with is protein partners including cell surface amino oxidases such as vascular adhesion protein 1 (VAP-1) (also known as AOC3) a known non-classical inflammation-inducible endothelial molecule (39). The over-activity of MAOB can result in the observed PEA concentration decrease and the major changes observed in the polyamines as well as related amino acids in the delirium-prone patients. Pro-inflammatory stimuli, including cytokines lead to MAO-dependent increases in reactive oxygen species causing mitochondrial dysfunction (14). At the same time, interference with MAOB activity in subjects with overactive MAOB can lead to the side-effects observed in MAOB inhibition as well as SARS-CoV-2 infection. Metabolomics analysis and symptom similarity of pre-operative delirium prone patients as well as COVID-19 patients indicates the potential dysfunction of MAOB. Computational modeling defines a mechanism by which the spike protein directly binds to MAOB thereby interfering with its normal function and particularly affecting patients with increased MAOB expressions. Further analysis is currently under way to explore in greater detail the role of MAOB in delirium and SARS-CoV-2 infection.

## Materials and Methods

### Samples and experimental analyses

Preoperative blood and CSF samples were collected within an observational cohort study of patients aged 65 years without a diagnosis of dementia presenting for planned hip and knee replacements prior to the SARS-CoV-2 pandemic (approved by Office for Research Ethics Committee for Northern Ireland (REC ref: 10/NIR01/5)). The study methodology has been described in detail previously (8). The study population had a mean age of 74.4 years and 57% were female. The incidence of postoperative delirium was 14%. Paired CSF and blood metabolomics analysis was undertaken for 54 age and gender matched patients where half of the patients experienced post-operative delirium and half did not (control). Metabolomic analysis was undertaken where sufficient CSF was available for age and gender matched delirium and control participants.

### Metabolomics

The analysis of metabolic profiles of CSF samples from the nested case-control postoperative delirium cohort has previously been published (10). In this investigation the corresponding blood plasma samples of each of the same patients was examined by an identical kit-based methodology. Quantitative metabolomic profiling was performed using the Biocrates AbsoluteIDQ p180 (BIOCRATES, Life Science AG, Innsbruck, Austria) using a Xevo TQ-MS triple-quadrupole mass spectrometer (Waters Corporation, Milford, USA) as previously described (41). Briefly, this comprised of two general methods: UPLC (I-Class, Waters Corporation, UK) reversed-phase (Waters ACQUITY UPLC BEH C18 2.1 × 50 mm, 1.7 μm; UK) with multiple reaction monitoring (MRM), and flow injection analysis (FIA) also operating in MRM mode. Metabolite concentrations were calculated and expressed as micromolar (µM).

### SARS-CoV-2 metabolomics dataset

Plasma metabolomics profiles of SARS-CoV-2 patients was published and made available by Shen et al. (11). The provided dataset includes metabolomic analysis of serum samples from patients with mildmild as well as, severe COVID-19 as well as control subjects (with number of patient equal to mild 28, severe 37 and control 28). Briefly, authors used the ultra-performance liquid chromatography/tandem mass spectrometry (UPLC-MS/MS) for untargeted metabolomics of serum samples providing identification and quantification of 941 metabolites including 36 drugs and their metabolites. Details of methodology and validation are provided in the original publication (11).

### Data analysis

Different machine learning and statistical methods running under Matlab 2020 (Matworks Inc), Orange 3.25 (42), TMeV (43) and Python - Jupyter Notebook were used for the analysis of metabolomics data. Principal component analysis (PCA) and 2-class Partial Least Squares-Discriminant Analysis (PLS-DA) model investigation was performed using jupyter notebook on data that was log10 transformed with selection to have an under 20% standard deviation and fewer than 10% of missing values in measurements for all samples. Selection of major features in different groups was performed using Significant Analysis of Microarrays (SAM) (18). Machine learning analysis was performed using Python and Matlab with the results of Python Random Forest classification with SHAP (SHapley Additive exPlanations) algorithm shown to explain the output of machine learning (44). A comparison of the protein structures of MAOB and ACE2 bonded to SARS-CoV-2 Spike protein are analyzed using UCSF Chimera, Schrodinger software package (Schrodinger Inc.) and PDB Data Bank structure analysis methodologies including jFATCAT (27). Protein X-ray structures were obtained from [https://www.rcsb.org/](https://www.rcsb.org/) and included for MAOB protein PDB ID 1GOS (45) and ACE2 bonded to SARS-CoV-2 Spike protein PDB ID: 6M0J.

## Conclusion

Significant differences were observed in a number of mono- and polyamines which led us to investigate in some detail the relationship between the observed changes in operative delirium and delirium caused by SARS-CoV-2 and to propose a hypothetical relationship between monoamine oxidase and SARS-CoV-2 spike protein. Further computational and experimental analysis of the relationship between delirium and SARS-CoV-2 and the possibility for spike-protein binding to monoamine oxidase is currently underway.

## Data Availability

Data requests can be send to the authors.

## Acknowledgments

Many people contributed to the successful completion of this study. In particular we gratefully acknowledge the support of the anaesthetists, surgeons, theatre and ward staff of Musgrave Park Hospital; Mr John Conlon of the Centre for Experimental Medicine, QUB for technical laboratory support; Professor Chris Patterson of the Centre for Public Health for statistical support; Dr Rebecca Cairns and Ms Lauren Anderson for data inputting; Mrs Hazel Johnston and Mrs Eilish Armstrong for neuropsychology training; the Cheung family for support via the Siew Keok Chin scholarship. Alzheimer’s Research UK (Metabolomic Analyses funded by Network Centre Pump Priming Grant). Cohort Study Funded by: Siew Keok Chin Scholarship (Philanthropic Funding). Belfast Arthroplasty Research Trust and Belfast Trust Charitable Funds.

MCC would like to thank Drs. Adrian Culf, Mary-Ellen Harper and Louis Borgeat for their reading of the manuscript and National Research Council for supporting this research.

## Footnotes

*   **Classification** BIOLOGICAL SCIENCES - Applied Biological Sciences; Biochemistry; Biophysics and Computational Biology;

*   Received June 16, 2020.
*   Revision received June 16, 2020.
*   Accepted June 19, 2020.


*   © 2020, Posted by Cold Spring Harbor Laboratory

This pre-print is available under a Creative Commons License (Attribution-NonCommercial-NoDerivs 4.0 International), CC BY-NC-ND 4.0, as described at [http://creativecommons.org/licenses/by-nc-nd/4.0/](http://creativecommons.org/licenses/by-nc-nd/4.0/)

## References

1.  1. H. A. Rothan,  S. N. Byrareddy, The epidemiology and pathogenesis of coronavirus disease (COVID-19) outbreak. Journal of Autoimmunity, 109:102433, (2020).
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.jaut.2020.102433&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2020%2F06%2F19%2F2020.06.16.20128660.atom) 

2.  2. L. Mao,  H. Jin,  M. Wang et al. Neurologic manifestations of hospitalized patients with coronavirus disease 2019 in Wuhan, China. JAMA Neurol (2020).
    
    

3.  3. J. Helms,  S. Kremer,  H. Merdji et al. Neurologic features in severe SARS-CoV-2 infection. N Engl J Med (2020).
    
    

4.  4. J. Lan,  J. Ge,  J. Yu, et al. Structure of the SARS-CoV-2 spike receptor-binding domain bound to the ACE2 receptor. Nature. (2020).
    
    

5.  5. J. Shang,  Y. Wan,  C. Luo, et al. (2020). Cell entry mechanisms of SARS-CoV-2. Proceedings of the National Academy of Sciences 1–8 (2020).
    
    

6.  6. A.B. Gussow,  N. Auslander,  G. Faure, et al. Genomic determinants of pathogenicity in SARS-CoV-2 and other human coronaviruses. BioRxiv, 2020.04.05.026450. (2020).
    
    

7.  7. T.G. Fong,  D. Davis,  M. Growdon,  A. Albuquerque,  S.K. Inouye, The Interface of Delirium and Dementia in Older Persons. Lancet Neurol., 21(10), 3240–3248 (2015).
    
    

8.  8. E.L. Cunningham,  B. McGuinness,  D.F. McAuley Df, et al. CSF Beta-amyloid 1-42 Concentration Predicts Delirium Following Elective Arthroplasty Surgery in an Observational Cohort Study. Ann Surg 269:1200–1205. (2019).
    
    

9.  9. A. Cleare,  C.M. Pariante, A.H. Young, A. H., et al. (2015). Evidence-based guidelines for treating depressive disorders with antidepressants: A revision of the 2008 British Association for Psychopharmacology guidelines. Journal of Psychopharmacology 29, (2008)
    
    

10. 10. X. Pan,  E.L. Cunningham,  A.P. Passmore, et al. Cerebrospinal Fluid Spermidine, Glutamine and Putrescine Predict Postoperative Delirium Following Elective Orthopaedic Surgery. Scientific Reports, 9(1), 1–9 (2019).
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1038/s41598-019-39057-w&link_type=DOI) 

11. 11. B. Shen,  X. Yi, Y. Sun, Y., et al. Proteomic and Metabolomic Characterization of COVID-19 Patient Sera. SSRN Electronic Journal, 1–14, (2020).
    
    

12. 12. R.C. Becker, COVID-19 update: Covid-19-associated coagulopathy. Journal of Thrombosis and Thrombolysis, 0123456789 (2020).
    
    

13. 13. W.A. Alkeridy,  I. Almaghlouth,  R. Alrashed, et al. A Unique Presentation of Delirium in a Patient with Otherwise Asymptomatic COVID-19. Journal of the American Geriatrics Society, 1–3. (2020)
    
    

14. 14. S. Deshwal,  M. Forkink,  C.H. Hu, et al. Monoamine oxidase-dependent endoplasmic reticulum-mitochondria dysfunction and mast cell degranulation lead to adverse cardiac remodeling in diabetes. Cell Death and Differentiation, 25(9), 1671–1685. (2018).
    
    

15. 15. O. Leiter,  T.L. Walker, Platelets in Neurodegenerative Conditions—Friend or Foe? Frontiers in Immunology, 11(May), 1–14, (2020).
    
    

16. 16. A.W.K. Yeung,  M.G. Georgieva,  A.G. Atanasov,  N.T. Tzvetkov, Monoamine oxidases (MAOs) as privileged molecular targets in neuroscience: Research literature analysis. Frontiers in Molecular Neuroscience, 12: 1–12, (2019).
    
    

17. 17. M. Ketzef,  G. Spigolon,  Y. Johansson, et al. Dopamine Depletion Impairs Bilateral Sensory Processing in the Striatum in a Pathway-Dependent Manner. Neuron, 94, 855-865.e5. (2017).
    
    

18. 18. V.G. Tusher,  R. Tibshirani,  G. Chu, Significance analysis of microarrays applied to the ionizing radiation response. Proceedings of the National Academy of Sciences, 98(9), 5116–5121. (2001).
    
    [Abstract/FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6NDoicG5hcyI7czo1OiJyZXNpZCI7czo5OiI5OC85LzUxMTYiO3M6NDoiYXRvbSI7czo1MDoiL21lZHJ4aXYvZWFybHkvMjAyMC8wNi8xOS8yMDIwLjA2LjE2LjIwMTI4NjYwLmF0b20iO31zOjg6ImZyYWdtZW50IjtzOjA6IiI7fQ==) 

19. 19. M. Cuperlovic-Culf,  D. Ferguson,  A. Culf,  P. Morin, M. Touaibia, M. 1H NMR metabolomics analysis of glioblastoma subtypes: correlation between metabolomics and gene expression characteristics. The Journal of Biological Chemistry, 287(24), 20164–20175, (2012).
    
    [Abstract/FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6MzoiamJjIjtzOjU6InJlc2lkIjtzOjEyOiIyODcvMjQvMjAxNjQiO3M6NDoiYXRvbSI7czo1MDoiL21lZHJ4aXYvZWFybHkvMjAyMC8wNi8xOS8yMDIwLjA2LjE2LjIwMTI4NjYwLmF0b20iO31zOjg6ImZyYWdtZW50IjtzOjA6IiI7fQ==) 

20. 20. S. Schedin-Weiss,  M. Inoue,  L. Hromadkova, et al. Monoamine oxidase B is elevated in Alzheimer disease neurons, is associated with γ-secretase and regulates neuronal amyloid β-peptide levels. Alzheimer’s Research and Therapy, 9(1), 1–19, (2017).
    
    

21. 21. M.A. Sharpe, D.S. Baskin, D. S. Monoamine oxidase B levels are highly expressed in human gliomas and are correlated with the expression of HiF-1α and with transcription factors Sp1 and Sp3. Oncotarget, 7(3), 3379–3393, (2016).
    
    

22. 22. H.G. Brunner,  M. Nelen,  X.O. Breakefield,  H.H. Ropers and  B. A. Vanoost, Abnormal behavior associated with a point mutation in the structural gene for monoamine oxidase A. Science, 262, 578–580 (1993)
    
    [Abstract/FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6Mzoic2NpIjtzOjU6InJlc2lkIjtzOjEyOiIyNjIvNTEzMy81NzgiO3M6NDoiYXRvbSI7czo1MDoiL21lZHJ4aXYvZWFybHkvMjAyMC8wNi8xOS8yMDIwLjA2LjE2LjIwMTI4NjYwLmF0b20iO31zOjg6ImZyYWdtZW50IjtzOjA6IiI7fQ==) 

23. 23. J.C. Shih,  K. Chen and  M. J. Ridd, Monoamine Oxidase: From Genes to Behavior Annu. Rev. Neurosci. 22, 197–217 (1999)
    
    

24. 24. S. Gandhi and  A. Y. Abramov, Oxidative Stress in Neurodegenerative Diseases and Ageing Oxid. Med. Cell. Longevity, 428010 (2012).
    
    

25. 25. M. Naoi,  W. Maruyama,  K. Inaba-Hasegawa,  Y. Akao, Modulation of monoamine oxidase (MAO) expression in neuropsychiatric disorders: genetic and environmental factors involved in type A MAO expression Int. Rev. Neurobiol., 100, 85–106, (2011).
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=21971004&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2020%2F06%2F19%2F2020.06.16.20128660.atom) 

26. 26. D. Pchejetski,  O. Kunduzova,  A. Dayon, et al. Oxidative stress-dependent sphingosine kinase-1 inhibition mediates monoamine oxidase A-associated cardiac cell apoptosis. Circulation Research, 100(1), 41–49, (2007).
    
    [Abstract/FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6MTA6ImNpcmNyZXNhaGEiO3M6NToicmVzaWQiO3M6ODoiMTAwLzEvNDEiO3M6NDoiYXRvbSI7czo1MDoiL21lZHJ4aXYvZWFybHkvMjAyMC8wNi8xOS8yMDIwLjA2LjE2LjIwMTI4NjYwLmF0b20iO31zOjg6ImZyYWdtZW50IjtzOjA6IiI7fQ==) 

27. 27. Y. Ye &  A. Godzik Flexible structure alignment by chaining aligned fragment pairs allowing twists. Bioinformatics 19 suppl. 2. ii246–ii255, (2003)
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1093/bioinformatics/btg1086&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=14534198&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2020%2F06%2F19%2F2020.06.16.20128660.atom) 
    
    [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000207434300029&link_type=ISI) 

28. 28. J.R. Maldonado, Neuropathogenesis of delirium: Review of current etiologic theories and common pathways. American Journal of Geriatric Psychiatry, 21(12), 1190–1222. (2013).
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.jagp.2013.09.005&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=24206937&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2020%2F06%2F19%2F2020.06.16.20128660.atom) 
    
    [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000330363800003&link_type=ISI) 

29. 29. Z. Xie, G.M. Miller, G. M. β-phenylethylamine alters monoamine transporter function via trace amine-associated receptor 1: Implication for modulatory roles of trace amines in brain. Journal of Pharmacology and Experimental Therapeutics, 325(2), 617–628. (2008).
    
    [Abstract/FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6NDoianBldCI7czo1OiJyZXNpZCI7czo5OiIzMjUvMi82MTciO3M6NDoiYXRvbSI7czo1MDoiL21lZHJ4aXYvZWFybHkvMjAyMC8wNi8xOS8yMDIwLjA2LjE2LjIwMTI4NjYwLmF0b20iO31zOjg6ImZyYWdtZW50IjtzOjA6IiI7fQ==) 

30. 30. S.S. Kang,  E.H. Ahn,  Z. Zhang, et al. α-Synuclein stimulation of monoamine oxidase-B and legumain protease mediates the pathology of Parkinson’s disease. The EMBO Journal, 37(12), 1–19, (2018).
    
    [Abstract/FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6NzoiZW1ib2pubCI7czo1OiJyZXNpZCI7czo2OiIzNy8xLzEiO3M6NDoiYXRvbSI7czo1MDoiL21lZHJ4aXYvZWFybHkvMjAyMC8wNi8xOS8yMDIwLjA2LjE2LjIwMTI4NjYwLmF0b20iO31zOjg6ImZyYWdtZW50IjtzOjA6IiI7fQ==) 

31. 31. J.H. Park,  Y.H. Ju,  J.W. Choi, et al. Newly developed reversible MAO-B inhibitor circumvents the shortcomings of irreversible inhibitors in Alzheimer’s disease. Science Advances, 5(3), 1–13, (2019).
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1126/sciadv.aax2950&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=31535024&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2020%2F06%2F19%2F2020.06.16.20128660.atom) 

32. 32. J.G. Fiedorowicz,  K.L. Swartz,. The role of monoamine oxidase inhibitors in current psychiatric practice. Journal of Psychiatric Practice, 10(4), 239–248, (2004).
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1097/00131746-200407000-00005&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=15552546&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2020%2F06%2F19%2F2020.06.16.20128660.atom) 

33. 33.R.M. Tackley, B. Tregaskis Fatal disseminated intravascular coagulation following a monoamine oxidase inhibitor/tricyclic interaction. Anaesthesia, 42(7), 760–763, (1987).
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=3631476&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2020%2F06%2F19%2F2020.06.16.20128660.atom) 
    
    [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=A1987J003300015&link_type=ISI) 

34. 34. M. Thorp,  D. Toombs,  B. Harmon, Monoamine oxidase inhibitor overdose. Western Journal of Medicine, 166(4), 275–277, (1997).
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=9168689&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2020%2F06%2F19%2F2020.06.16.20128660.atom) 

35. 35. M Sandler,  M A Reveley, V Glover Human platelet monoamine oxidase activity in health and disease: a review. J Clin Pathol. 34(3): 292–302, (1981).
    
    [Abstract/FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6OToiamNsaW5wYXRoIjtzOjU6InJlc2lkIjtzOjg6IjM0LzMvMjkyIjtzOjQ6ImF0b20iO3M6NTA6Ii9tZWRyeGl2L2Vhcmx5LzIwMjAvMDYvMTkvMjAyMC4wNi4xNi4yMDEyODY2MC5hdG9tIjt9czo4OiJmcmFnbWVudCI7czowOiIiO30=) 

36. 36. E. Cassol,  V. Misra, s, Morgello et al. Altered Monoamine and Acylcarnitine Metabolites in HIV-Positive and HIV-Negative Subjects With Depression. J Acquir Immune Defic Syndr.;69(1):18–28. (2015).
    
    

37. 37. A. Gulati,  C. Pomeranz,  Z. Qamar,  S Thomas, et al. A Comprehensive Review of Manifestations of Novel Coronaviruses in the Context of Deadly COVID-19 Global Pandemic. Am J Med Sci. May 11,(2020).
    
    

38. 38. P. Xu,  Q. Zhou, J. Xu, J. Mechanism of thrombocytopenia in COVID-19 patients. Annals of Hematology, 1205–1208. (2020).
    
    

39. 39. M. Salmi, S. Jalkanen, S. VAP-1: An adhesin and an enzyme. Trends in Immunology, 22(4), 211–216. (2001).
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/S1471-4906(01)01870-1&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=11274927&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2020%2F06%2F19%2F2020.06.16.20128660.atom) 
    
    [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000169936200010&link_type=ISI) 

40. 40. E.L. Cunningham,  T. Mawhinney,  D. Beverland et al. Age Ageing. Sep 1;46(5):779–786, (2017).
    
    

41. 41. X. Pan,  M. Nasaruddin,  M. Bin, et al. Alzheimer’s disease-like pathology has transient effects on the brain and blood metabolome. Neurobiology of Aging, 38, 151–163, (2016).
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=26827653&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2020%2F06%2F19%2F2020.06.16.20128660.atom) 

42. 42. J. Demsar,  T. Curk,  A. Erjavec et al. Orange: Data Mining Toolbox in Python, Journal of Machine Learning Research 14(Aug): 2349−2353, (2013).
    
    

43. 43. A. Saeed,  V. Sharov, J. White J, et al. TM4: a free, open-source system for microarray data management and analysis. Biotechniques. 34(2):374–8, (2003).
    
    [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/S0076-6879(06)11009-5&link_type=DOI) 
    
    [PubMed](http://medrxiv.org/lookup/external-ref?access_num=12613259&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2020%2F06%2F19%2F2020.06.16.20128660.atom) 
    
    [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000180892500018&link_type=ISI) 

44. 44. S.M. Lundberg,  S. Lee, A Unified Approach to Interpreting Model Predictions, Advances in Neural Information Processing Systems 30, 4765—4774, (2017).
    
    

45. 45. C. Binda,  P. Newton-Vinson,  F. Hubalek, et al. Human Monoamine Oxidase B Nat Struct Biol 9: 22 (2001)
    
    

46. 46. J. Robinson, et al.,. An atlas of human metabolism. Sci. Signal. 13, eaaz1482, (2020).
    
    [Abstract/FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6ODoic2lndHJhbnMiO3M6NToicmVzaWQiO3M6MTU6IjEzLzYyNC9lYWF6MTQ4MiI7czo0OiJhdG9tIjtzOjUwOiIvbWVkcnhpdi9lYXJseS8yMDIwLzA2LzE5LzIwMjAuMDYuMTYuMjAxMjg2NjAuYXRvbSI7fXM6ODoiZnJhZ21lbnQiO3M6MDoiIjt9) 

47. 47. B C Mcloughlin,  A Miles,  T E Webb,  P Knopp, et al. Functional and cognitive outcomes after COVID-19 delirium medRxiv 2020.06.07.20115188; (2020).