Identification of perioperative neurocognitive dysfunction biomarkers in cerebrospinal fluid with quantitative proteomic approach in patients undergoing transurethral resection of prostate with combined spinal and epidural analgesia

Background:

This study aimed to identify predictive biomarkers of perioperative neurocognitive dysfunction (PND) in cerebrospinal fluid of elderly male patients undergoing elective transurethral resection of prostate, using an isobaric tags for relative and absolute quantitative-based quantitative proteomic approach.

Methods:

Patients were evaluated with Mini Mental State Examination at −1 and+3 days of operation. Presence of PND was determined with Z-score method. Patients characteristics and quantitative cerebrospinal fluid proteomes detected with isobaric tags for relative and absolute quantitative-were compared between PND and non-PND patients. Gene ontology and Kyoto Encyclopedia of Genes and Genomes analysis were performed to identify pathways potentially involved in PND.

Result:

A total of 229 patients were included in the study and 32 were diagnosed with PND (incidence 14.4%). The age, incidence of hypertension, and diabetes of PND patients were significantly higher than non-PND patients (P < .05). There were 85 differentially expressed proteins identified, among which High Mobility Group Box 1, prostaglandin D synthase, and matrix metalloproteinase inhibitor were considered to be promising candidates as they might play important roles in pathophysiology of PND.

Conclusion:

Proteomic approach identified potential biomarkers for predicting the occurrence of PND. These findings need to be validated in further studies.

1. Introduction

Perioperative neurocognitive dysfunction (PND) is increasingly impacting economy and society as human life expectancy and number of operations in elderly patients increases.^[1] around 15% of patients over 60-years old shows objectively measured cognitive decline as consequence of anesthesia and surgery.^[2] Characterized by cognitive decline, anxiety, confusion, personality and behavior changes, mental decline, and impaired memory function,^[3] PND was formally defined by multidisciplinary experts in 2018 standing for changes in cognitive function associated with anesthesia and surgery. A systematic review by Paredes et al^[4] reported that among 6477 patients, 756 were diagnosed with PND, resulting in an incidence of 11.7% in 19 studies using individual analyses of neurocognitive tests. The incidence of PND after non-cardiac major surgery was 7% to 26% and even higher than patients over 60-years-old, while the incidence of PND in patients undergoing cardia surgery may reach 30% to 80% within a few weeks after surgery and 10% to 60% in 3 to 6 months after surgery.^[5] PND is associated with postoperative complications and worsened prognosis including chronic and neuropathic pain, increased hospitalization time/cost, and delayed recovery,^[6,7] resulting in poor quality of life, increased financial burden, and increased mortality.^[8] Therefore, the diagnosis and treatment of PND has become a problem to be solved.

PND is a complex process that may involve in many causal and inducing factors. Risk factors for PND have been identified in multiple studies including increase of age, prolonged duration of anesthesia or surgery, low education levels, postoperative delirium, and use of sedatives. Meanwhile, it was reported that silent stroke,^[9,10] hypertension, diabetes, and cardiac disease are important preoperative predictors for PND.^[11,12] Previous studies showed that the pathophysiological mechanisms of PND were related to central nervous system degeneration, neuroinflammation response, blood–brain barrier (BBB) dysfunction, pain, circadian rhythm sleep disorders, and change in neuro transmitters. It was also reported that excess activation of astrocytes, neuroinflammation, hippocampal neuron aging, and BBB dysfunction in PND were associated with change in plasma biomarker levels.^[13–16] Given the fact that central nervous system dysfunction can significantly affect the composition and concentration of proteins in cerebrospinal fluid (CSF), CSF protein profile may represent a candidate predictor for PND. Several proteins such as TNF-α, S-100β, tau protein, and certain interleukins might potentially serve as predictors of PND.^[17–20] However, considering the complex nature of PND, one single biomarker is unlikely to be suitable for patients under different conditions, a proteomic approach may be more advantageous. The aim of this study was to identify differential expressed proteins (DEPs) with proteomic approach between PND and non-PND patients who underwent transurethral resection of the prostate (TURP) under combined spinal and epidural analgesia to identify potential predictors of PND, which was done through isobaric tags for relative and absolute quantitative (iTRAQ) technology. iTRAQ was a quantitative proteomic approach developed on recent years^[21] and a powerful tool in not only characterizing protein expression profile but also biomarker identification. Thus far few researches regarding the proteomic profile of PND have been reported.

2. Methods

2.1. Setting and patients

This study was carried out at Hospital (T.C.M) Affiliated to Southwest Medical University, Luzhou, Sichuan Province, China and the study protocol was approved by the ethics committee of Southwestern Medical University of China (Protocol number: 20190306-05). The study was registered at the Chinese Clinical Trial Registry (Study ID ChiCTR2000028836). Written informed consent was obtained from all participants. From March 2019 to December 2019, patients receiving elective TURP at the center were consecutively included if the following inclusion criteria were met: Age > 65; American Society of Anesthesiologists Classification (ASA Class) 1 to 3; Preoperative Mini Mental State Examination (MMSE) greater than minimum score of corresponding level of education (illiteracy ≥ 17, primary school culture ≥ 20, middle school and above ≥ 23). If the following exclusion criteria are met, they will not be included continuously: drug abuse history; abnormal coagulation function; organ failure; hemorrhagic or ischemic lesion in central nervous system; Parkinson disease or Alzheimer disease; viral hepatitis; HIV infection tuberculosis, or other chronic infectious disease. Patients with major depressive or maniac disorder, other neuropsychiatric condition or insufficient eyesight and/or hearing to follow instructions were also excluded.

2.2. Randomization and masking

Cognitive function of patients was assessed with MMSE by a senior psychiatrist 1 day before and 3 days after surgery. Presence of PND was determined with Z-score based method, where 20 healthy people (age > 65) who did not receive surgery and had normal cognitive function during the same period was used as reference group. Z score was defined as:

$$Z=\frac{\Delta X-\Delta X_{\text{Reference}}}{SD{(\Delta X)}_{\text{Reference}}}$$

where $\Delta X$ stands for the difference of 2 MMSE measurements before and after surgery in TURP patients, $\Delta X_{\text{Reference}}$ stands for the difference of 2 MMSE measurements of reference group individuals (this value was subtracted to adjust for practice effect), and $SD{(\Delta X)}_{\text{Reference}}$ stands for the standard deviation of difference of 2 MMSE measurements of reference group individuals. PND was diagnosed if $\left|\text{Z}\right|\text{ >1}\text{.96}$ (Rasmussen et al). According to Z-score, TURP patients were classified as PND group and non-PND group.

During surgery, combined spinal and epidural analgesia was performed on all participants without preanesthetic medication. L2-L3 or L3-L4 space was selected for subarachnoid puncture and 1 mL CSF sample was subsequently taken and stored at −80°C. Ropivacaine 10 to 15 mg was then injected and level of anesthesia was maintained at T6 to T8. Sedation was not applied during the operation. After surgery, patient-controlled intravenous analgesia was placed.

For proteomic analysis with iTRAQ technology, CSF samples from PND and non-PND group patients were used. Albumin and IgG were depleted with Sigma-Aldrich ProteoPrep® Blue Albumin & IgG Depletion Kit. The protein samples were then digested with trypsin and labeled with 8-plex iTRAQ kit according to the manufacturer’s protocol. The labeled protein samples were analyzed with LC-MS/MS analysis with Thermo Fisher Q Exactive HF. Raw data were searched in Uniprot Homo Sapiens database. DEPs were classified by gene ontology (GO) annotation based on biological process (BP), cellular component, and molecular function (MF). GO annotation proteome was derived from the UniProt-GOA database (www.http://www.ebi.ac.uk/GOA/). Kyoto Encyclopedia of Genes and Genomes (KEGG) database was used to annotate protein pathway. Two-tailed Fisher exact tests were applied to test the GO, KEGG pathway and Domain enrichment of the differential expression protein for all identified proteins. Correction for multiple hypothesis testing was carried out using standard false discovery rate control methods and a corrected P-value < .05 was considered to be statistically significant. Expression-based clustering and functional enrichment-based clustering for different protein groups were used to explore potential relationships between different protein groups at special protein function. Cluster membership were visualized through a heat map with the “pheatmap” R-package.

2.3. Statistical methods

For baseline characteristics between PND and non-PND group patients, one-way ANOVA was used compare age, body mass index (BMI), operation time, intravenous fluid infusion volume, hemorrhage volume (calculated with rinse volume, hemoglobin concentration of rinse and preoperative hemoglobin concentration), MMSE score, and Z-score values. Chi-square test was used to compare ASA classification, hypertension, and diabetes incidence between groups. Quantitative data were presented with mean ± standard error. P values < .05 were considered to be of statistical significance.

3. Results

In total, 229 patients were included in the study. Seven patients were excluded from the analysis since 1 patient had severe intraoperative bleeding, 2 patients had anaphylaxis during anesthesia, and 4 patients were lost to follow-up. Finally, 222 patients were eligible for statistical analysis whose MMSE measurements and CSF samples were taken. Among these patients 32 developed PND (incidence 14.41%). Between PND and non-PND groups, no statistical significance was observed in ASA grading, BMI, MMSE score, operation time, hemorrhage volume, and intravenous fluid infusion volume, while PND group had older age (P < .001), higher incidence of hypertension, and diabetes (P < .05) (Table 1 and Fig. 1).

Table 1.

Baseline characteristics, intraoperative characteristics and perioperative cognitive function measurements of PND and non-PND group patients.

Feature	PND group (n = 32)	Non-PND group (n = 190)	P value
Age (yr)	75.09 ± 4.86	72.35 ± 3.65	<.001
ASA classification
I	8 (25%)	50 (26.3%)	.95
II	15 (46.9%)	83 (43.7%)
III	9 (28.1%)	57 (30%)
BMI (cm/kg2)	22.41 ± 2.26	21.25 ± 3.21	.52
Comorbidity
Hypertension	10 (31.3%)	30 (15.8%)	.035
Diabetes	8 (25%)	22 (11.6%)	.040
Operation time (min)	87.28 ± 18.09	93.1 ± 26.18	.36
IV fluid infusion volume (mL)	753.12 ± 148.07	759.82 ± 126.28	.78
Hemorrhage volume (mL)	86.62 ± 35.70	86.38 ± 45.14	.98
Preoperative MMSE score	26.13 ± 1.77	26.16 ± 1.64	.91
Postoperative MMSE score	20.16 ± 1.19	24.51 ± 1.52	<.001
Z-score	3.24 ± 0.37	1.16 ± 0.14	<.001

Data were presented as mean ± standard error.

ASA = American Society of Anesthesiologists, BMI = body mass index, IV = intravenous, MMSE = Mini-Mental State Examination, PND = perioperative neurocognitive dysfunction, postoperative MMSE score = MMSE measured 3 days after surgery, preoperative MMSE score = MMSE measured 1 day before surgery.

Figure 1.

CONSORT 2010 flow diagram.

Two CSF samples from PND group patients (PND-1 and PND-2) and 2 CSF samples from non-PND (non-PND) group patients were randomly selected for mass spectrometry analysis to identify DEPs. A total of 470 proteins, of which 451 were quantifiable, were identified (error recovery rate < 1%). Proteins that had elevated or reduced median ratio of more than 1.2 times and P value ≤ .05 in both Group 1 (PND-1 compared to Non-PND) and Group 2 (PND-2 compared to non-PND) results were considered as DEPs. In Group 1, 121 DEPs were identified including 66 upregulated and 55 downregulated proteins (Fig. 2A). In Group 2, 125 DEPs were identified including 67 upregulated and 58 downregulated proteins (Fig. 2B). The combination of Group 1 and Group 2 results yielded 85 DEPs of which 40 were upregulated and 45 were downregulated. Table 2 showed a complete list of 85 identified DEPs, among which 8 proteins including High Mobility Group Box 1 (HMGB1), Apolipoprotein A-II (APOA), Apolipoprotein C-I (APOC), nucleophosmin (NPM), prostaglandin D synthase (PTGDS), Kallikrein-6 (KLK6), matrix metalloproteinase inhibitor (TIMP1), and transthyretin (TTR) were selected for further analysis (Table 3) and the heatmap of them was plotted (Fig. 3).

Figure 2.

(A) In Group 1 (PND-1 compared to non-PND), 121 DEPs were identified including 66 upregulated and 55 downregulated proteins. (B) In Group 2 (PND-1 compared to non-PND), 125 DEPs were identified including 67 upregulated and 58 down regulated proteins. PND-1/2: CSF sample randomly drawn from PND group. Non-PND: CSF sample randomly drawn from non-PND group. CSF = cerebrospinal fluid, DEPs = differential expressed proteins, PND = perioperative neurocognitive dysfunction.

Table 2.

A complete list showing all differential expressed protein identified with mass spectrometry between PND and non-PND cerebrospinal fluid samples.

Protein	Number	Accession	Regulated-stage	BLAST score	N75-vs-N79	N111-vs-N79	Description
P22392	1	P22392	Down	317	0.41	0.58	Nucleoside diphosphate kinase B
P61978	2	P61978	Down	940	0.48	0.61	Heterogeneous nuclear ribonucleoprotein K
O75541	3	O75541	Down	738	0.30	0.42	Zinc finger protein 821
Q8NGH9	4	Q8NGH9	Down		0.49	0.35	Olfactory receptor 52E4
P06748	5	P06748	Down	598	0.24	0.26	Nucleophosmin
B2RPK0	6	B2RPK0	Down	321	0.31	0.48	Putative high mobility group protein B1-like 1
Q9GZV4	7	Q9GZV4	Down	317	0.35	0.33	Eukaryotic translation initiation factor 5A-2
Q6NXT2	8	Q6NXT2	Down	258	0.48	0.52	Histone H3.3C
P02652	9	P02652	Down		0.88	0.80	Apolipoprotein A-II
P02654	10	P02654	Down		0.59	0.60	Apolipoprotein C-I
Q9UM44	11	Q9UM44	Down		0.15	0.17	HERV-H LTR-associating protein 2
Q14847	12	Q14847	Down	550	0.44	0.68	LIM and SH3 domain protein 1
P30101	13	P30101	Down	1044	0.32	0.38	Protein disulfide-isomerase A3
P27797	14	P27797	Down	839	0.38	0.39	Calreticulin
P19823	15	P19823	Down	173	0.74	0.77	Inter-alpha-trypsin inhibitor heavy chain H2
P11021	16	P11021	Down	1326	0.31	0.34	Endoplasmic reticulum chaperone BiP
P05062	17	P05062	Down	759	0.60	0.60	Fructose-bisphosphate aldolase B
O60814	18	O60814	Down	254	0.42	0.29	Histone H2B type 1-K
P01876	19	P01876	Down		0.72	0.74	Immunoglobulin heavy constant alpha 1
P08708	20	P08708	Down	276	0.35	0.71	40S ribosomal protein S17
P38646	21	P38646	Down	1392	0.34	0.40	Stress-70 protein, mitochondrial
P37198	22	P37198	Down	1025	0.32	0.42	Nuclear pore glycoprotein p62
P07737	23	P07737	Down	291	0.60	0.71	Profilin-1
P62913	24	P62913	Down	366	0.40	0.46	60S ribosomal protein L11
P10809	25	P10809	Down	1139	0.30	0.38	60 kDa heat shock protein, mitochondrial
P07237	26	P07237	Down	1037	0.39	0.49	Protein disulfide-isomerase
P58304	27	P58304	Down	453	0.26	0.29	Visual system homeobox 2
Q93088	28	Q93088	Down	839	0.46	0.60	Betaine--homocysteine S-methyltransferase 1
P02647	29	P02647	Down		0.76	0.69	Apolipoprotein A-I
P05787	30	P05787	Down		0.31	0.20	Keratin, type II cytoskeletal 8
P04114	31	P04114	Down	9381	0.67	0.69	Apolipoprotein B-100
P01861	32	P01861	Down		0.78	0.66	Immunoglobulin heavy constant gamma 4
P42765	33	P42765	Down	811	0.20	0.21	3-ketoacyl-CoA thiolase, mitochondrial
Q9NU22	34	Q9NU22	Down	4065	0.55	0.75	Midasin
Q8WWF5	35	Q8WWF5	Down	852	0.40	0.42	E3 ubiquitin-protein ligase ZNRF4
P60174	36	P60174	Down	511	0.27	0.29	Triosephosphate isomerase
Q32P51	37	Q32P51	Down	642	0.60	0.76	Heterogeneous nuclear ribonucleoprotein A1-like 2
P14927	38	P14927	Down	225	0.35	0.36	Cytochrome b-c1 complex subunit7
Q86XE5	39	Q86XE5	Down	247	0.30	0.29	4-hydroxy-2-oxoglutarate aldolase, mitochondrial
P23528	40	P23528	Down	335	0.51	0.49	Cofilin-1
Q96PQ6	41	Q96PQ6	Down	1242	0.43	0.42	Zinc finger protein 317
Q15493	42	Q15493	Down	625	0.34	0.36	Regucalcin
P31327	43	P31327	Down	3108	0.47	0.49	Carbamoyl-phosphate synthase [ammonia], mitochondrial
P62937	44	P62937	Down	339	0.40	0.43	Peptidyl-prolyl cis-trans isomerase A
P46779	45	P46779	Down	278	0.58	0.58	60S ribosomal protein L28
P02763	46	P02763	Up		1.49	1.68	Alpha-1-acid glycoprotein 1
P02766	47	P02766	Up	302	1.24	1.88	Transthyretin
P02042	48	P02042	Up	303	1.43	1.93	Hemoglobin subunit delta
Q6P587	49	Q6P587	Up	463	1.23	1.49	Acylpyruvase FAHD1, mitochondrial
P02748	50	P02748	Up		1.23	1.34	Complement component C9
Q5SYB0	51	Q5SYB0	Up	3243	1.74	2.10	FERM and PDZ domain-containing protein 1
Q12805	52	Q12805	Up	1019	1.33	1.35	EGF-containing fibulin-like extracellular matrix protein 1
Q16270	53	Q16270	Up		1.27	1.66	Insulin-like growth factor-binding protein 7
P10643	54	P10643	Up		1.38	1.36	Complement component C7
P69905	55	P69905	Up	286	1.48	2.33	Hemoglobin subunit alpha
Q13822	56	Q13822	Up	1775	1.31	1.53	Ectonucleotide pyrophosphatase/phosphodiesterase family member 2
P36222	57	P36222	Up	797	1.25	1.25	Chitinase-3-like protein 1
P07998	58	P07998	Up		1.28	1.33	Ribonuclease pancreatic
P51693	59	P51693	Up	1313	1.29	1.21	Amyloid-like protein 1
P05090	60	P05090	Up	393	1.29	1.61	Apolipoprotein D
P36955	61	P36955	Up	704	1.34	1.62	Pigment epithelium-derived factor
P36956	62	P36956	Up	2288	1.50	2.04	Sterol regulatory element-binding protein 1
P01009	63	P01009	Up	856	1.23	1.96	Alpha-1-antitrypsin
P61769	64	P61769	Up		1.78	2.20	Beta-2-microglobulin
Q9Y6V0	65	Q9Y6V0	Up		1.65	2.61	Protein piccolo
Q9GZU2	66	Q9GZU2	Up	3306	1.63	1.31	Paternally-expressed gene 3 protein
P23142	67	P23142	Up	1424	1.40	1.63	Fibulin-1
P06396	68	P06396	Up	1621	1.22	1.39	Gelsolin
Q8N960	69	Q8N960	Up		1.24	1.84	Centrosomal protein of 120 kDa
P08603	70	P08603	Up		1.20	1.24	Complement factor H
P10909	71	P10909	Up		1.44	1.49	Clusterin
Q15113	72	Q15113	Up	914	1.41	1.32	Procollagen C-endopeptidase enhancer 1
P01033	73	P01033	Up	432	1.62	1.67	Metalloproteinase inhibitor 1
P01034	74	P01034	Up		1.36	1.37	Cystatin-C
Q16610	75	Q16610	Up		1.33	1.26	Extracellular matrix protein 1
P68871	76	P68871	Up	301	1.65	2.49	Hemoglobin subunit beta
P00746	77	P00746	Up	457	1.45	1.29	Complement factor D
P02774	78	P02774	Up		1.31	1.42	Vitamin D-binding protein
Q8WXW3	79	Q8WXW3	Up		1.34	1.67	Progesterone-induced-blocking factor 1
P02675	80	P02675	Up	1030	1.37	1.65	Fibrinogen beta chain
P41222	81	P41222	Up	49.3	1.38	1.54	Prostaglandin-H2 D-isomerase
A0A1B0GW35	82	A0A1B0GW35	Up	74.3	1.71	1.49	Exocyst complex component 1-like
P17900	83	P17900	Up		1.37	1.29	Ganglioside GM2 activator
P08294	84	P08294	Up	491	1.38	1.30	Extracellular superoxide dismutase [Cu-Zn]
Q92876	85	Q92876	Up	508	1.28	1.45	Kallikrein-6

PND = perioperative neurocognitive dysfunction.

Table 3.

Information of 8 differential expressed proteins selected for further analysis.

Protein ID	Name	Status	Ratio (PND/non-PND)	P value
P06748	Nucleophosmin	Downregulate	0.25	.034
B2RPK0	High Mobility Group Box 1	Downregulate	0.395	.012
P02652	Apolipoprotein A-II	Downregulate	0.84	.016
P02654	Apolipoprotein C-I	Downregulate	0.595	.014
P02766	Transthyretin	Upregulate	1.56	.021
P41222	Prostaglandin-H2 D-isomerase	Upregulate	1.46	.028
Q92876	Kallikrein-6	Upregulate	1.365	.031
P01033	Matrix metalloproteinase inhibitor1	Upregulate	1.645	.044

PND = perioperative neurocognitive dysfunction.

Figure 3.

Heatmap of 8 differential expressed proteins selected for further analysis.

These selected DEPs were analyzed through GO annotation and classified according to their participation in BP (Fig. 4A), cellular component (Fig. 4B), and MF (Fig. 4C) to explore the potential function of the proteins. In upregulated proteins (TTR, PTGDS, KLK6, and TIMP1), BP involved included regulation of protein metabolic process, response to wounding, negative regulation of multicellular organismal process, single-multicellular organism process and regulation of proteolysis; CC involved included blood microparticle, extracellular matrix, perinuclear region of cytoplasm, HFE–transferrin receptor complex, apical cortex and keratin filament; MF involved included oxygen binding, oxygen transporter activity, peptidase regulator activity, structural molecule activity, transition metal ion binding, and heparin binding. In downregulated proteins (NPM, HMGB1, APOA, and APOC), relevant BP included carboxylic acid metabolic process, cellular modified amino acid metabolic process, alpha-amino acid metabolic process, oxoacid metabolic process, organic acid metabolic process, and small molecule biosynthetic process; relevant CC included focal adhesion, cell-substrate adherens junction, cell-substrate junction, extracellular exosome, extracellular vesicle, extracellular organelle, and extracellular region part; relevant MF included unfolded protein binding, transaminase activity, transferase activity, transferring nitrogenous groups, kynurenine aminotransferase activity, kynurenine oxoglutarate transaminase activity, and immunoglobulin receptor binding (Fig. 4). KEGG pathway analysis of DEPs revealed 6 significantly enriched pathways including Alzheimer disease related pathway MAP04610 and neuroactive ligand-receptor interactions related pathway MAP04080, as was shown in Table 4. The functions of involved pathways could be summarized as complement and coagulation cascade, malaria, neuroactive ligand/receptor interaction, endoplasmic reticulum protein processing, alcoholism, and fat digestion and absorption (Fig. 5). Results of domain enrichment analysis was shown in Figure 6.

Figure 4.

Functional annotation of 8 selected DEPs including HMGB1, APOA, APOC, NPM, PTGDS, KLK6, TIMP1, and TTR. (A) BP analysis of upregulated and downregulated DEPs. (B) CC analysis of upregulated and downregulated DEPs. (C) MF analysis of upregulated and downregulated DEPs. APOA = Apolipoprotein A-II, APOC = Apolipoprotein C-I, BP = biological process, CC = cell composition, DEP = differential expressed protein, HMGB1 = High Mobility Group Box 1, KLK6 = Kallikrein-6, MF = molecular function, NPM = nucleophosmin, PTGDS = prostaglandin D synthase, TIMP = matrix metalloproteinase inhibitor, TTR = transthyretin.

Table 4.

KEGG pathway analysis of 8 selected DEPs including HMGB1, APOA, APOC, NPM, PTGDS, KLK6, TIMP1, and TTR.

Pathway ID	Pathway_Name	Diff-mapping	Protein-Mapping	Diff-Num	Protein-Num	Fold-Enrichment	Fisher exact test P value	Mapping -Protein IDs
map04610	Complement and coagulation cascades	10	37	28	227	2.19	0.01	P10909; P05546; P00746; P01009; P02675; P08603; P01042; P10643; P00734; P02748
map05144	Malaria	2	3	28	227	5.40	0.04	P68871; P69905
map04080	Neuroactive ligand-receptor interaction	2	3	28	227	5.40	0.04	P00734; P07477
map04977	Vitamin digestion and absorption	3	3	65	227	3.49	0.02	P04114; P02647; P06727
map04141	Protein processing in endoplasmic reticulum	4	6	37	227	4.09	0.01	P30101; P27797; P11021; P07237
map05034	Alcoholism	3	5	37	227	3.68	0.03	O60814;Q6NXT2; P62805

APOA = apolipoprotein A-II, APOC = apolipoprotein C-I, DEP = differential expressed protein, HMGB1 = High Mobility Group Box 1, KEGG = Kyoto Encyclopedia of Genes and Genomes, KLK6 = Kallikrein-6, NPM = nucleophosmin, PTGDS = prostaglandin D synthase, TIMP = matrix metalloproteinase inhibitor, TTR = transthyretin.

Figure 5.

KEGG pathway analysis of upregulated (TTR, PTGDS, KLK6, and TIMP1) and downregulated (NPM, HMGB1, APOA, and APOC) DEPs. APOA = Apolipoprotein A-II, APOC = Apolipoprotein C-I, DEP = differential expressed protein, HMGB1 = High Mobility Group Box 1, KEGG = Kyoto Encyclopedia of Genes and Genomes, KLK6 = Kallikrein-6, NPM = nucleophosmin, PTGDS = prostaglandin D synthase, TIMP = matrix metalloproteinase inhibitor, TTR = transthyretin.

Figure 6.

Domain enrichment analysis of KEGG analysis showing DEPs were involved in complement and coagulation cascades. DEP = differential expressed proteins, KEGG = Kyoto Encyclopedia of Genes and Genomes.

4. Discussion

In this study, no statistically significant difference was found ASA grading, BMI, MMSE score between the PND and non-PND groups. A total of 85 DEPs were identified with iTRAQ-based quantitative proteomic approach, of which 8 DEPs were selected for further analysis including HMGB1, APOA, APOC, NPM (which were downregulated), PTGDS, KLK6, TIMP1, and TTR (which were upregulated). GO annotation and pathway analysis were also performed. The results suggested that PTGDS, HMGB1 and TIMP1 might play important roles in pathophysiology of PND.

Glutathione independent PTGDS is a platelet aggregation inhibitor involved in pain, sleep, and smooth muscle adduction/release.^[22,23] It is involved in multiple central nervous system functions including sedation, NREM sleep, and PGE2-induced heterotopia pain. PTGDS may have anti-apoptotic effect in oligodendrocytes and play an important role in the maturation and maintenance of the central nervous system and male reproductive system. There are 2 types of PTGDS: L-PGDS and H-PGDS. L-PGDS, also known as cerebral PGDS or glutathione independent PGDS, is secreted by the pia mater and arachnoid membrane into CSF, and is the second most important protein in human CSF after albumin. L-PGDS has different glycosylated structures in different body fluids. In CSF, L-PGDS has a unique complex oligosaccharide structure characterized by high fucoglycosylation, large amount of n-acetylglucosamine, N-acetylglucosamine, and galactose terminal. Under physiological conditions, L-PGDS can be detected in CSF, semen, ascites, amniotic fluid, urine, serum, and other different body fluids related to the physiological functions of the central nervous system and male reproductive system. L-PGDS is also one of the key rate-limiting enzymes that catalyze the synthesis of PGD2 by PGH2.^[24] Five subtypes of PTGDS have been found in CSF in people with no identifiable psychiatric or neurological disease. PTGDS is involved in multiple physiological functions including regulation of sleep cycles, body temperature, pituitary hormone, pain, etc., and altered PTGDS level in CSF may affect multiple BP.^[25] Experiments showed PTGDS was related to inflammatory response including brain inflammation and arthritis,^[26,27] suggesting PTGDS might be an important candidate of biomarkers for PND due to its close relation to relevant pathophysiological mechanisms.

HMGB1 is a highly conserved non-histone chromosomal binding protein present in the eukaryotic nucleus actively secreted by certain cell types (macrophages, etc) and passively secreted by dead or infected cells.^[28] In cerebral hemorrhage, tumor, epilepsy, trauma, or other secondary chronic inflammatory situation, HMGB1 can start immune response and expand inflammatory response by acting on specific receptors.^[29] Previous studies showed HMGB1 may be a key mediator of surgically induced cognitive decline and one of the reasons for change in BBB permeability after peripheral surgical trauma.^[30,31] In stroke model, HMGB1 induced activation of astrocytes, microvascular structures, changes in endothelial cell function and increased BBB permeability. In endothelial cells, recombinant human HMGB1 induces pro-inflammatory response and activation of microglia, resulting in increased vascular permeability, cell swelling and destruction of the blood-brain barrier, which is closely.^[32–34] Meanwhile, HMGB1 may interact with a variety of receptors to mediate the release of pro-inflammatory cytokines, leading to up-regulation of pro-inflammatory cytokines including TNF-a, IL-b in mice during anesthesia and surgical trauma.^[35,36] However, in this study, decreased expression level of HMGB1 was observed in CSF, suggesting the underlying mechanism needs further investigation in future study.

Matrix metalloproteinases (MMPs) are a group of Zn-dependent endopeptidases capable of degrading and remodeling the extracellular matrix and increasing BBB permeability, which are key to the PND induced by systemic inflammation.^[37] Studies showed that various exogenous injury factors or cerebral ischemia-reperfusion injury may induce inflammatory response and production of free radicals, which could activate MMP9, MMP3 and other MMPs to degrade the basal membrane components and extracellular matrix of BBB, causing damage to the BBB. TIMP is a specific tissue inhibitor of MMP that can inhibit the degradation of extracellular matrix by MMP through binding to the progenitor and activated form of MMPs. Four members of TIMP family have been identified. In PND mouse model, imbalance between MMPs and TIMP were observed 6 hours postoperatively, leading to decreased expression of zonula occludens-1 and increased permeability of BBB, making the brain accessible to proinflammatory mediators at 24 hours after surgery and eventually resulted in inflammatory response and cognitive dysfunction, suggesting the imbalance between MMPs and MIMP may be a mechanism of PND, which was also supported by the results of this study.

In conclusion, this study utilized an iTRAQ-based quantitative proteomic approach to identify DEPs in CSF samples obtained from PND patients and proposed potential biomarkers to predict the occurrence of PND. Meanwhile, it was important to note that the study was only preliminary and exploratory as it was limited by relatively small sample size in quantitative proteomic analysis and the characteristics of participants. Further researches are needed to validate the finding in larger scale and examine if the conclusion can be generalized to other patients and/or PND after other types of surgery. Secondly, the operation selection of this study is “transurethral resection of prostate,” and the scope is narrow, and the follow-up scope can be expanded to explore the comparison of markers of different surgical types, and follow-up animal experiments should be carried out on the markers selected at present to further explore their role in PND.

Author contributions

Conceptualization: Ying Zhang, Qing Liu.

Data curation: Tian-Yan Luo.

Formal analysis: Tian-Yan Luo.

Software: Tian-Yan Luo.

Supervision: Wei Zhou.

Validation: Wei Zhou.

Writing – original draft: Gui-Fang Xiang.