CSF PGRN May be Associated With Postoperative Delirium After Knee Replacement in Elderly Patients: The PNDABLE Study

Bin Wang Qingdao Municipal Hospital Group https://orcid.org/0000-0002-1384-5964 Jie Xiu Sun Qingdao Municipal Hospital Group He Tao Dalian Medical University Yuan Xi Deng Dalian Medical University Nan Ya Lin Weifang Medical College: Weifang Medical University Hao fang Liu Qingdao Municipal Hospital Group Xu Lin Qingdao Municipal Hospital Group Rui Dong Nanjing Medical University Qiu Yu Liu Qingdao Municipal Hospital Group Yanlin Bi (  yanlinbi68@sina.cn ) Qingdao Municipal Hospital https://orcid.org/0000-0002-9006-976X


Background
Postoperative delirium (POD) represents a serious complication following anesthesia and surgical procedures for patients undergoing surgical intervention (1). POD is characterized by temporary or permanent cognitive decline, memory impairment, deterioration in language comprehension and social adaptation ability, and POD particularly affects elderly people (> 65 years) (2). POD can lead to increased mortality, prolonged hospitalization, other complications such as Alzheimer's disease, and higher treatment costs (3). Despite the prevalence and clinical importance of POD, its pathophysiology is poorly understood and no reliable biomarkers have been reported in previous studies.
PGRN, a multifunctional secretory protein, is a neurotrophic growth factor. It is encoded by a single gene on chromosome 17q21. Composed of 593 amino acids, it is rich in cysteine. And it has a molecular weight of 68.5KDa. The precursor proteins are hydrolyzed by extracellular proteases such as elastin into smaller peptide fragments called GRNs or epithelial proteins (4). These segments range in size from 6 KDa to 25KDa. PGRN exists in a large number of tissues and organs of the human body, mainly in epithelial tissues (5). On the body surface, PGRN is mainly involved in post-injury repair and in ammatory response (6). PGRN produces vascular endothelial factors in skin cancer cells (7). In the central nervous system, PGRN gene mainly exists in speci c neurons, including microglial cells, cerebellar Purkinje cells and hippocampal pyramidal neurons, which have the functions of neurotrophy, prolongation of axons, promotion of neuron survival and the proliferation of neural stem cells (8). Studies have shown that PGRN expression is signi cantly increased during neuroin ammation (9), and PGRN growth in microglia cells may play an important role in brain injury, neuroin ammation and neurodegeneration (10). The increase in PGRN expression in neurodegenerative diseases may be a self-protective mechanism to prevent cell damage. Some studies have found that PGRN protein is closely related to changes in cognitive function (11). POD and AD both belong to neurodegenerative diseases. Therefore, to the best of our knowledge, no previous study has studied the association between CSF PGRN and POD. Therefore, we speculated that patients with higher preoperative CSF PGRN levels were prone to POD. The main objective of this study was to investigate the associations of preoperative CSF PGRN concentration with POD occurrence and CSF core biomarkers for POD, including Aβ 1−42 , Aβ 1−40 , p-tau and T-tau.

PNDABLE study
The Perioperative Neurocognitive Disorder and Biomarker Lifestyle Study (PNDABLE) is intended to explore the pathogenesis, risk factors and biomarkers of perioperative neurocognitive disorders in the northern Chinese Han population. PNDABLE is aimed to identify lifestyle factors that may affect the risk of PND in the non-demented northern Chinese Han population in order to provide a basis for disease prevention and early diagnosis. This study has important scienti c and practical values for establishing the standardized model of early diagnosis and prevention for PND in China. Informed consent was obtained from all the included patients before we extracted preoperative cerebrospinal uid and blood of the patients. This study has been registered in the Chinese Clinical Trial Registry (clinical registration number ChiCTR2000033439) and approved by the Ethics Committee of Qingdao Municipal Hospital.

Participants
The Han Chinese patients undergoing unilateral total knee arthroplasty (no gender limitations, aged 65 ~ 90, weight 50-80 kg, ASA ~ ) combined with epidural anesthesia were enrolled in the PNDABLE study at Qingdao Municipal Hospital (East Hospital) from June 2020 to November 2020. The exclusion criteria include: (1) Preoperative CM-MMSE score < 24 points; (2) A history of neurological and mental diseases such as Alzheimer's disease, Parkinson's disease, and cerebrovascular accident, etc.; (3) Drug or psychotropic substance abuse, as well as long-term use of steroid drugs and hormone drugs; (4) preoperative -hepatic encephalopathy; (5) Recent major surgery; (6) Severe visual and hearing impairments; (7) Abnormal coagulation function before surgery.
A total of 600 cognitively normal participants from PNDABLE had available information on covariates. We excluded 25 participants who had no information about CM-MMSE, 5 participants without available CSF PGRN data, 17 participants who had no CSF biomarker data or had data outside four standard deviations (SD) of the mean, and 8 participants whose surgeries were suspended. Finally, 545 participants were included in this analysis and they were divided into two groups according to whether POD occurred or not: POD group and non-POD group. POD cases and non-POD controls were frequency-matched (1:1) on ve variables using incidence density sampling. Speci cally, one non-delirium control was randomly selected for each POD case from the source population according to the ve matched variables, including age, diagnosis, American Society of Anesthesiologist' (ASA) physical status, duration of surgery, and intraoperative blood loss. These variables were listed in the European Society of Anesthesiology evidence-based and consensus-based guideline on POD and were considered to be risk factors for POD after hip fracture surgery (Aldecoa et al., 2017). A patient recruitment owchart is shown in Fig. 1.
The participants did not receive preoperative medications, and they were instructed not to drink for 6 h and not to eat for 8 h before surgery. After entering the operating room, we routinely monitored ECG, SpO 2 and NBP, opened vein access and extracted 3 ml of whole venous blood. All patients underwent combined spinal-epidural block, with the space between lumbar 3-4 spinous processes (L3-4) as the puncture site. After successful puncture, 2 ml of cerebrospinal uid was extracted from the subarachnoid space, followed by injection of 2-2.5 ml ropivacaine (0.66%) for about 30 s. After anesthesia, the sensory level was controlled below the T8 level. During the surgery, oxygen was inhaled via mask at 5L/min to maintain blood pressure within ± 80% of the baseline value. If intraoperative NBP < 90 mmHg (1mmHg = 0.133 kPa) or it decreased by more than 20% of the baseline value, ephedrine 5 mg was injected intravenously. If HR < 50 beats/min, atropine 0.5 mg was injected intravenously. Intravenous patient-controlled analgesia (butorphanol 0.1 mg/ml + tropisetron 50 g/ml, diluted with normal saline to a total volume of 100 ml) was used in acute postoperative pain management. After the operation, the patient was sent to the anesthesia resuscitation room (PACU). If no abnormalities were found during a 30-minute observation period, then the patient could return to the ward with low-ow oxygen and continuous monitoring of vital signs.
We interviewed all the patients the day before surgery and collected their baseline data, including age, gender, body mass index (BMI), ASA physical status, years of education, as well as CM-MMSE, CAM and Memorial Delirium Assessment Scale (MDAS) scores. Other information including comorbidities, past medical history, fracture classi cation, types of anesthesia and surgery, and time from injury to operation were also collected according to the patients' medical records. All the history collection, physical evaluation and cognitive assessment related to dementia were conducted by neurologists.
CSF core biomarker and CSF PGRN measurements CSF samples were processed immediately within 2 h after standard lumbar puncture. Each sample was centrifuged at 2000 × g for 10 min, and CSF samples were separated and stored in an enzyme-free EP (Eppendorf) tube (AXYGEN; PCR-02-C) at − 80 °C for further use in the subsequent steps of this study. The samples were subjected to a maximum of two freeze-thaw cycles.
CSF PGRN and core biomarkers were measured by ELISA using the microplate reader (X) (Thermo Scienti c Multiskan MK3). CSF PGRN measurements were done with ELISA kits (Human PGRN SimpleStep ELISA kit; BioVendor, no. RMEE103R) and CSF core biomarker measurements were done with other ELISA kits (INNOTEST; FUJIREBIO). All ELISA measurements were performed by experienced technicians in strict accordance with the manufacturer's instructions. They were blinded to the clinical information. The samples and standards were measured in duplicate, and the means of duplicates were used for the statistical analyses. All the antibodies and plates were from a single lot to exclude variability between batches. Moreover, the within-batch CV was < 5% and the inter-batch CV was < 15%.

Neuropsychological tests
The Chinese-Modi ed Mini-Mental State Examination (CM-MMSE) was completed by neurologists 1d before surgery to assess the preoperative cognitive status and record relevant medical history. Patients whose MMSE scores < 23 points were excluded. Participants received interview preoperatively and in PACU, on the rst, second, third and seventh (or before discharge) postoperative days. The assessment of delirium was performed in PACU, on the rst, second, third and seventh days (or before discharge) after surgery between 9:00 am and 11:00 am by neurologists. We used the visual analog scale (VAS) score of 0-10 (lower scores indicating lower levels of pain (12)) to assess pain at the same time. POD was de ned by the Confusion Assessment Method (CAM) (13), and POD severity was measured using the Memorial Delirium Assessment Scale (MDAS) (14). The Chinese versions of CAM and MDAS have been proven to have good reliability and validity in the Chinese elderly population [15,16]. Therefore, CAMpositive and MDA-positive patients postoperatively in PACU and on the rst, second, third and seventh days (or before discharge) were recorded.
CSF PGRN didn't follow a normal distribution as assessed by Kolmogorov-Smirnov test (P < 0.001) and visual inspection of the Q-Q plot (Fig. 1S). Therefore, they were log-transformed to obtain a normal distribution. All the statistical analyses described in this study are performed on the log10-transformed values. We performed the analysis after excluding outliers (de ned as 4 SD below or above the group mean) in order to exclude the in uence of extreme values. Two independent-samples' t tests were used for the comparisons between POD and NPOD groups. We used the Correlation analysis to explore whether CSF PGRN is related to CAM score and MDAS score. Given the different trends of PGRN at different ages in the biomarker framework, we applied a one-way ANCOVA followed by Bonferroni post hoc analyses.
We also studied the associations between CSF PGRN and the CSF core biomarkers for POD, using a multiple linear regression adjusted for age, gender, years of education, and APOE ε4 carrier status. The analyses were performed in the total sample and then in subgroups strati ed by age, gender, years of education and APOE ε4 carrier status.
ROC curve analysis was used to evaluate the clinical diagnostic value of PGRN in POD. Statistical signi cance was set at P < 0.05. SPSS statistical software, version 21.0 (SPSS, Inc. Chicago, IL, USA), and GraphPad Prism software, version 6.01 (GraphPad Software, Inc., La Jolla, CA, USA), were used for data analysis.

Participant characteristics
A total of 600 Han Chinese patients over the age of 65 who underwent unilateral total knee arthroplasty were included in the PNDABLE study from January 2018 to January 2020. The reasons for dropping out are shown in Fig. 1. And 545 patients (n = 545) remained for analyses. We found the incidence of POD was 9.7% (n = 53 of the 545 patients) via our postoperative assessments. Another 53 non-POD patients were also enrolled in this study (Fig. 1).

Differences in CSF PGRN level between different subgroups strati ed by biomarkers
The associations between CSF PGRN and CSF core biomarkers for POD were tested in linear regression models adjusted for age, gender, years of education and APOE ε4 carrier status. In the whole sample of subjects (n = 659), increased CSF PGRN was associated with lower levels of  (Fig. 6).
We then calculated the ratios between CSF amyloid and tau biomarkers, and found no associations of CSF PGRN with CSF Aβ 1−40 /T-tau or Aβ 1−40 /p-tau (Table 2). Outliers were excluded in our analyses, but we obtained similar results when those were included. These ndings indicate that higher CSF PGRN correlates with lower levels of Aβ and higher levels of tau.

Receiver operating characteristic (ROC) curve analysis of PGRN in CSF
The ROC curve analysis of PGRN showed that PGRN concentration had high diagnostic value for POD, with all the AUC greater than 0.5 and close to 1.0. (Table 3, Fig. 7).

Discussion
The incidence of POD in our study was 9.7%, which was consistent with the previous results of 3.6-41% [17]. For example, previous studies have shown that the incidence of POD after the total knee and hip replacement under spine anesthesia is 20% [18]. There is still a great deal of controversy about the pathogenesis of POD. At present, there are several assumptions including cholinergic theory, in ammatory reaction and stress-response theory. There are many risk factors for POD, such as advanced age, preexisting cognitive decline, blood loss and blood transfusion, anesthetic medications, as well as postoperative pain, etc. [19]. Therefore, our study adopted CAM and MDAS to improve the accuracy of our assessments of POD.
In the present study, we combined biomarker-based classi cation with age to assess changes in CSF PGRN (a marker for microglial activity) in POD patients. The application of this classi cation system enabled us to explore the associations between microglial in ammatory response and the pathophysiology of POD (including Aβ pathology, tau pathology, and neurodegeneration). Our study showed that CSF PGRN levels did change dynamically with aging.
Aβ pathology (de ned as low CSF Aβ 1−42 and Aβ 1−40 ) was associated with an increase in CSF PGRN, while tau pathology or neurodegeneration was associated with elevated CSF PGRN. This seems to con rm the potential role of microglial in ammatory response in the pathogenesis of POD. Our results support the hypothesis that Aβ deposition occurs independently of the in ammatory process, but the type and extent of the in ammatory response to Aβ deposition in the brain may trigger or affect subsequent neurodegeneration. Another piece of evidence is that immunotherapy against amyloid can reduce downstream neurodegeneration, a process that may be mediated by changes in microglial activation [20].
In this study, we explored the associations between CSF PGRN and CSF biomarkers for POD to further clarify the pathogenesis of POD and provide theoretical basis for early warning and intervention of POD. Our results showed that in the entire data set and in POD, CSF PGRN was positively correlated with T-tau, P-tau, Aβ 42 /T-tau and Aβ 42 / ptau, as well as negatively correlated with Aβ 1−42 or Aβ 1−40 , further suggesting that POD was related to reactive microglia proliferation. The underlying mechanism of CSF PGRN remains to be investigated throughout the disease. In NPOD subjects, the correlations of PGRN with Aβ 40 and T-tau disappeared, while the correlation of PGRN with ptau remained. These ndings suggest that CSF PGRN may indeed be associated with neuronal injury. It is also suggested that increased CSF PGRN in NPOD patients may be a protective response to mild neuronal injury.
Progranulin (PGRN) is a multifunctional growth factor expressed in a variety of tissues and involved in many physiological and pathological processes [21]. It is widely expressed in various cells of the body. Some studies have found that PGRN is highly expressed in neurons and microglia cells [22]. Little was previously known about the role of PGRN in the nervous system, but since the discovery of PGRN genetic polymorphisms, the number of studies on the role of PGRN in the brain has increased rapidly. Studies have found that PGRN can activate microglia cells and stimulate them to engulf the toxic Aβ around them, which exerts neuroprotective effects [23]. Other studies have found that the content of PGRN in microglia cells which are around Aβ deposits increases in mice [24]. Neuro brillary tangles are one of the main pathological features of AD, which are closely related to two major proteins --Tau protein and CDK5 [25]. Tau protein is found throughout the nervous system, and its hyperphosphorylation is one of the early cytoskeletal changes during the formation of NFT [26]. Generally speaking, Tau protein is modi ed by 2-3 phosphate groups. The phosphorylation and dephosphorylation of tau protein maintain a dynamic balance, maintaining the stability of cytoskeleton [27]. In the pathological state of AD, Tau protein has 9-10 phosphate groups, leading to its hyperphosphorylation and the formation of NFT [28]. Hyperphosphorylated tau protein loses its original functions and cannot promote microtubule focusing and maintain cytoskeleton stability [29]. Moreover, hyperphosphorylated Tau protein competes with normal Tau protein to bind microtubules, resulting in an increase in hyperphosphorylated Tau protein and a decrease in normal Tau protein. CDK5, also known as Tau kinase, mainly regulates tau phosphorylation. CDK5 has been shown to be closely associated with neurodegenerative diseases [30]. In the pathological process of AD, the increased expression of CDK5 not only directly aggravates Tau hyperphosphorylation, but also plays a role in regulating phosphatases or kinases of Tau protein. Studies have found that when PGRN is upregulated, it activates central cyclin-dependent kinase (CDK), which leads to reduced clearance of toxic Aβ and oxidative stress [31]. Neuro brillary tangles and neuronal loss caused by Tau hyperphosphorylation lead to cognitive dysfunction [32].
The ROC curve analysis showed that PGRN concentrations had the greatest diagnostic value. Therefore, high PGRN concentrations can predict the occurrence and development of POD before surgery. Increased CSF PGRN and its effects have been observed in the brains of AD patients and AD model mice. Increased expression of PGRN in microglia cells around amyloid plaques is a self-protective mechanism to prevent cell damage, which offers prospects for the application of CSF PGRN as a biomarker for patients with cognitive dysfunction. Therefore, it is the future direction of our research to replicate our ndings in animal experiments and explore the relevant mechanisms.
Our investigation had two limitations. Firstly, this is a cross-sectional study that limits any conclusions about disease progression. Therefore, results should be replicated in subjects with longitudinal data to analyze whether CSF PGRN levels are associated with disease progression. Second, cerebrospinal uid collection is an invasive procedure.
Monitoring the concentration of PGRN in the plasma of patients will make clinical examination more convenient.
This project will monitor the progression of the disease by measuring the changes of PGRN concentration in the peripheral blood through large-scale clinical studies.

Conclusion
In conclusion, this study is based on an independent cohort. The results indicate that the occurrence and development of cognitive dysfunction in elderly patients after unilateral total knee arthroplasty may be related to the increased expression of PGRN in CSF, and the concentration of PGRN in CSF increases with age. Aβ pathology is associated with a decrease in CSF PGRN in the absence of tau deposition and neurodegeneration, whereas tau pathology and neurodegeneration are associated with an increase in CSF PGRN. Future studies should use CSF biomarkers to further explore the biological mechanisms underlying POD. Availability of data and materials

Abbreviations
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.
Ethics approval and consent to participate The present study followed the recommendations of the National Institute of Health guidelines for the care and use of laboratory animals and obtained approval from the Clinical Trial Ethics Committee of Qingdao Municipal Hospital, Qingdao, China.

Consent for publication
Not applicable.

Figure 1
Flow diagram showed selection of eligible patients and the enrollment process.  Associations of CSF PGRN and CSF core biomarkers. Scatter plots represent the associations of CSF PGRN with CSF core biomarkers: Aβ1-42, Aβ1-40, T-tau, p-tau, Aβ42/p-tau and Aβ42 / T-tau in NPOD groups . The normalized regression coe cients (β) and P values computed by multiple linear regression after adjustment for age, gender, educational level, and APOE ε4 carrier status are shown.

Figure 6
Associations of CSF PGRN and CSF core biomarkers. Scatter plots represent the associations of CSF PGRN with CSF core biomarkers: Aβ1-42, Aβ1-40, T-tau, p-tau, Aβ42/p-tau and Aβ42 / T-tau in POD patient groups. The normalized regression coe cients (β) and P values computed by multiple linear regression after adjustment for age, gender, educational level, and APOE ε4 carrier status are shown.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.