Introduction

Delayed neurocognitive recovery (dNCR), previously termed as postoperative cognitive dysfunction (POCD), is defined as a decline in cognitive function, including memory, information processing, and executive function, up to 30 days after surgery [1]. dNCR is associated with higher risk for impaired quality of life [2], longer term cognitive decline [3], and mortality within 1 year after surgery [4]. The incidence of dNCR varies by patient population, and it is even more frequent in elderly patients undergoing total joint arthroplasty (TJA) [5,6,7,8]. The pathophysiology of dNCR has not been fully understood, and no effective precautions and treatments are available currently.

Neuroinflammation is involved in the development of dNCR. Peripheral inflammatory responses to anesthesia and surgery disrupt the blood–brain barrier (BBB) and activate microglial function, leading to subsequent neuroinflammation [9]. Several animal studies have described that activation of cholinergic anti-inflammatory pathway (CAIP) may attenuate neuroinflammation and ameliorate POCD [10,11,12]. The CAIP refers to the binding of ACh to the α7nAChR of macrophages, which ultimately inhibits the release of inflammatory cytokines [13]. Cholinesterase hydrolyzes and inactivates ACh, thereby inhibiting the CAIP [14]. In animal studies, cholinesterase inhibitors reduced neuroinflammation and improved POCD [15, 16]. In human, a recent study also showed that higher peripheral AChE activity in patients was likely associated with postoperative delirium (POD) [17]. Thus, restraining inflammatory response by activating CAIP may reduce the occurrence of dNCR in elderly patients.

Vagus nerve stimulation (VNS) is a novel neuromodulation therapy that has been successfully applied in refractory epilepsy [18], depression [19], and migraine headaches [20]. The transauricular VNS (taVNS) that stimulates the auricular branch of the VN in a completely non-invasive manner improves cholinergic activity and activates the CAIP [21]. Recent data also proved that taVNS decreased the activity of cholinesterase and surgery-induced acute inflammatory responses [22, 23]. In addition, an animal experiment also suggested that the taVNS inhibited neuroinflammation and alleviated postoperative memory impairment in aged rats [24]. However, taVNS has not been studied in the prevention of postoperative cognitive decline in elderly patients.

The aim of this study was to investigate the effect of taVNS on dNCR in elderly patients undergoing TJA and to explore the mechanism of this intervention. We hypothesize that taVNS could reduce the incidence of dNCR by activating CAIP.

Methods

This prospective, randomized, double-blind, sham-controlled trial was conducted at the Third Hospital of Hebei Medical University, approved by the Institutional Review Board of the Third Hospital of Hebei Medical University (No. 2021-004-1), and registered in Chinese Clinical Trial Registry (ChiCTR2100044905). Informed consent was signed by each participant after full explanation of the procedure.

Participants

This study consecutively recruited elderly patients who underwent TJA in the Third Hospital of Hebei Medical University from April 2021 to October 2021. The main inclusion criteria were 65 years of age or older, American Society of Anesthesiologists (ASA) physical status I–III, and Mini-Mental State Examination (MMSE) score 24 and above. The exclusion criteria were history of schizophrenia and Parkinson’s disease; recent use of cholinergic and anticholinergic drugs, and hormones; suffering from inflammatory conditions, intake of anti-inflammatory drugs; communication issues such as serious hearing or visual impairment; an unwillingness to complete the experimental procedures.

Another 30 non-surgical matched control subjects who were primarily family members of surgical patients were enrolled to eliminate the practice effect [25] of repeated neuropsychological tests. The control group met the above inclusion and exclusion criteria except for surgery-related items. Non-surgical control subjects completed neuropsychological tests in the same manner as the surgery group.

Randomization and blinding

The consecutively recruited patients were divided into two groups according to a random number table: taVNS or sham stimulation (SS). The taVNS and SS were administered to the left ear by a commercial transcutaneous electrical nerve stimulation unit (Roscoe TENS 7000). For taVNS, a custom-developed stimulating electrode was placed on the cymba conchae and an available clip design behind the earlobe. For SS, the electrode clips were placed on the ear lobe and the tail of the helix (Fig. 1). The cymba conchae of the ear contains the highest density of auricular vagus nerve projections (100%), while the earlobe and the tail of the helix of the ear are free of cutaneous vagal innervation [26]. Stimulation parameters were set at a modulation mode with a frequency of 10 Hz, a pulse width of 300 μs, and an amplitude of the maximum amount that the subject could tolerate without pain. The stimulation started at 1 h prior to the anesthetic induction to the end of surgery. The electrodes were placed by the nurse and covered with opaque earmuffs. Patients and investigators were blinded to research group assignment.

Fig. 1
figure 1

Design of the study. a Experimental flow of the study. b Stimulation site of transauricular vagus nerve stimulation (taVNS) and Sham Stimulation (SS) on left ear. c Stimulation mode

Full size image

Anesthesia and perioperative management

Intraoperative monitoring included blood pressure, pulse oxygen saturation, electrocardiogram, bispectral index, and end-tidal CO2 (ETCO2). Anesthesia was induced intravenously with midazolam, sufentanil, propofol, and cisatracurium, followed by placement of a laryngeal mask airway. Anesthesia was maintained with intravenous propofol infusion and remifentanil infusion. Depth of anesthesia was maintained by controlling BIS 40 to 60. Palonosetron hydrochloride of 0.25 mg was administrated before the end of surgery to prevent postoperative nausea and vomiting (PONV). The use of hormones, nonsteroidal anti-inflammatory drugs, and anticholinesterase drugs was not allowed during the perioperative period.

Blood sample collection

Venous blood specimens were collected before surgery and on the first day after surgery. Serum was centrifuged for 15 min at 3000 g and saved frozen at − 80 °C. The activity of serum acetylcholinesterase (AChE) (A024-1-1) and butyrylcholinesterase (BChE) (A025-1-1) was assayed by a commercial enzyme immunoassay kit (Nanjing Jiancheng Bioengineering Institute). The concentrations of tumor necrosis factor-α (TNF-α) (A18210642, MultiSciences, China), interleukin-1β (IL-1β) (101B10151, MultiSciences, China), interleukin-6 (IL-6) (A10610651, MultiSciences, China), high-mobility group protein 1 (HMGB1) (CSBE08223h, CUSABIO, USA), and S100β (CSBE08065h, CUSABIO, USA) were quantified with enzyme-linked immunosorbent assay (ELISA) kit according to the manufacturer’s instructions.

Neuropsychological assessment

All patients were assessed by the Mini-Mental State Examination before surgery. Each participant underwent neuropsychological testing (using parallel forms of each test to reduce practice effects) at two time points: up to 3 days before surgery and at 1 week after surgery. Neuropsychological tests were included based on the protocol in the International Study of Postoperative Cognitive Dysfunction 2 (ISPOCD2) [27]: (1) Rey Auditory Verbal Learning Test (RAVLT)-immediate recall: to assess memory; (2) Trail making test (TMT): as a test of executive function; (3) Digit symbol substitution test (DSST): to assess processing speed; (4) Stroop Color Word Test (SCWT), part 3: as a test of attention.

Statistical analysis

Sample size calculation

The incidence of dNCR in elderly patients was approximately 35% after non-cardiac surgery [4, 28]. We hypothesized that the incidence of dNCR would decrease from 35 to 10% after taVNS. With a significance of 0.05 and a strength of 0.9, the minimum sample size required was 57 subjects per group or a total of 114 subjects (PASS 15.0). Considering a 10% abscission rate, the sample size was 126.

Outcome analysis

The incidence of dNCR at 1 week after surgery was defined as a decline in performance on neuropsychological assessment beyond natural variation. We defined dNCR as: a combined Z-score\(>\)1.96. For each patient, the Z-score for each neuropsychological test was calculated by this equation: Z = (\(\Delta\)X − \(\Delta\)X control) / SD(\(\Delta\)X control). \(\Delta\)X was the difference between postoperative and preoperative scores of surgical patients. The average practice effect \(\Delta\)X control was the mean of the difference between before and after the test interval for non-surgical control subjects. SD(\(\Delta\)X control) was the standard deviation of \(\Delta\)X control. The sign should be adjusted to ensure that a positive Z-score corresponds to a deterioration in all tests. The combined Z-score in an individual patient was calculated as the sum of Z-scores of the four tests divided by the standard deviation for this sum of Z-scores in the non-surgical control subjects. For specific methods, please refer to the corresponding literature [29].

Categorical variables were analyzed using Chi-square test or Fisher’s exact test. Continuous variables were tested with Student’s t test or Mann–Whitney U test. Intraoperative data and safety data were analyzed in the intent-to-treat population. The primary endpoint was analyzed in patients who completed 1 week neuropsychological tests. We used R software version 4.0.3 for above analyses.

Results

Subject characteristics

In total, 140 patients were enrolled from April 1, 2021, to October 12, 2021. Of the 140 eligible patients, 124 were randomized and finally analyzed (intention-to-treat analysis). Among 124 patients, 3 patients did not receive the assigned intervention (stimulation was terminated early in 2 patients, and 1 patient received combined intravenous-inhalation anesthesia). Of the 124 patients, 60 patients in the taVNS group and 59 patients in the SS group completed the neuropsychological tests (Fig. 2).

Fig. 2
figure 2

Flowchart of the study. ITT intent-to-treat. NT-tests neuropsychological tests

Full size image

Demographic and intraoperative characteristics did not differ significantly between the taVNS and SS groups (Table 1). The controls and surgical patients were matched for age, sex, BMI, education level, and MMSE score (Supplementary Table S1).

Table 1 Baseline characteristics
Full size table

Outcome analysis

Of all patients enrolled, 119 surgical patients (taVNS group, n = 60; SS group, n = 59) and 30 non-surgical control subjects completed 1 week neuropsychological tests. The mean scores for the neuropsychological tests and change scores (Z-score) corrected for learning effect are presented in Supplementary Table S2. The individual combined Z-score and the incidence of dNCR after surgery are shown in Fig. 3. The incidence of dNCR was significantly lower in the taVNS group (6 of 60 patients [10%]) than in the SS group (16 of 59 patients [27.1%]) (P = 0.016) (Fig. 3).

Fig. 3
figure 3

Scatter plot of combined Z-score of the taVNS and the SS groups. The dashed line represents combined Z-score = 1.96. The scatter inside the box is combined Z-score > 1.96. The incidence of delayed neurocognitive recovery was significantly lower in the taVNS group than the SS group (6/60 patients [10%] vs. 16/59 patients [27.1%], χ2 = 5.785, P = 0.016), analyzed by Pearson’s χ2 test

Full size image

There were no significant differences in cholinesterase activity and cytokines levels between the two groups at baseline. The postoperative AChE and BChE activity decreased significantly in the taVNS group compared to that preoperatively, but no significant change was found in the SS group (between-group comparison, P = 0.011 in AChE and P < 0.001 in BChE; Table 2). A marked inflammatory response occurred after TJA. All inflammatory factors in the two groups were significantly higher after surgery than those before surgery (P < 0.05). Compared with the SS group, taVNS significantly inhibited the increase in IL-6 (P = 0.007), HMGB1 (P = 0.012), and S100β (P = 0.01) induced by TJA. The magnitude of increase in TNF-α levels by TJA did not show any difference between the two groups (Table 2). IL-1β in a number of serum samples were below the limit of detection and no assessments could be made.

Table 2 Cholinesterase activity and cytokine
Full size table

Safety analysis

taVNS was well tolerated and did not cause adverse reactions such as pain, itching, and headache. There was no significant difference in the occurrence of bradycardia, hypotension, tachycardia, hypertension, and POVN between the two groups (Table 3).

Table 3 Safety outcomes
Full size table

Discussion

Our study found that taVNS during perioperative period could significantly reduce the incidence of dNCR at 1 week after TJA in elderly patients. The underlying mechanism may be related to the inhibition of inflammatory factors production and the reduction of AChE and BChE activity.

In this study, the incidence of dNCR in elderly adults undergoing TJA was 27.1%, which was consistent with the previous studies on postoperative cognitive decline in elderly patients undergoing orthopedic surgery (9.1–29%) [30, 31]. The taVNS significantly reduced the incidence of dNCR (10% vs. 27.1%). A recent animal experiment showed that the taVNS alleviated postoperative memory impairment in aged rats by inhibiting neuroinflammation [24]. Blood inflammatory factors that may cause neuroinflammation are closely related to POCD [32]. Therefore, we investigated serum levels of inflammatory markers.

The invasive nature of TJA activated an immune response and triggered the inflammatory cascade [33]. TNF-α, IL-6, and HMGB1 were significantly increased after TJA in this study. In addition, our results indicated that acute taVNS (approximately 3 h) significantly inhibited the surgery-induced increase of IL-6 and HMGB1 at 24 h after surgery, which was in accordance with the anti-inflammatory properties of the vagus nerve [34]. The activation of the anti-inflammatory response requires only a brief VNS stimulus and persists for more than 24 h [35]. The fMRI evidence in humans showed that the activation of all brain persisted throughout the 11 min after 7 min taVNS [36]. The taVNS may be followed by sustainable regeneration mechanisms which establish the sensory information flow to the brain [37]. Thus, the decrease of IL-6 and HMGB1 at 24 h after surgery may be attributed to the sustainable activation of VN and the initial inhibition of inflammatory cascade. IL-6 [38] and HMGB1 [39] disrupt BBB and contribute to subsequent neuroinflammation, resulting in elevated serum levels of S100β, a biomarker of cerebral or extra-cerebral neuron damage [40]. Thus, our study revealed that taVNS reduced the serum level of S100β, but did not decrease the level of TNF-α, which might be related to blood collection time and cytokine peak time [41].

Acetylcholine (ACh), as a main inflammation regulator in taVNS, can bind to α7 nicotinic acetylcholine receptor of macrophages and inhibit the production of pro-inflammatory cytokines (IL-6 and HMGB1) [42]. ACh is extremely unstable and difficult to use for clinical measurements, so the detection of hydrolyzing enzymes can be used as an indirect measure of Ach [43]. AChE and BChE hydrolyze and inactivate ACh [44]. The increase in the activity of enzymes AChE and BChE leads to reduced levels of ACh and systemic inflammatory response [45]. Byung et al. reported that VNS inhibited the production of pro-inflammatory cytokines by suppressing AChE in rats [46]. Our study showed that taVNS significantly reduced the postoperative activity of AChE and BChE as compared with the SS group. Our results support the hypothesis that AChE and BChE in serum antagonize CAIP at macrophage level and promote the systemic inflammatory response [47]. In addition, cholinesterase inhibitors also enhance vagus nerve activity to alleviate inflammation [48]. Therefore, the anti-inflammatory effect of taVNS may be related to the inhibition of cholinesterase activity.

Recently, several clinical studies reported the relationship between peripheral cholinesterase activity and POD. Muller et al. [17] reported higher peripheral AChE activity in patients with POD than in those without POD, possibly because higher AChE activity led to rapid degradation of ACh to impair cognition. However, Lin et al. [49] and Guenther et al. [50] showed low peripheral cholinesterase activity as a risk factor for POD. In their studies, in the absence of external intervention or cholinergic drugs’ treatment, low cholinesterase activity represented cholinergic hypofunction. This may be because low cholinesterase activity is associated with negative feedback of cholinergic deficiency [51]. Our results suggested that reducing peripheral cholinesterase activity might be beneficial for dNCR. VNS has been proved to reduce the expression of AChE and increase the levels of ACh in animal studies [46, 52, 53]. In this study, the decreased cholinesterase activity by taVNS represented increased levels of ACh and enhanced function of cholinergic system. Therefore, the above studies all showed that peripheral cholinesterase activity is linked to postoperative cognition, and cholinergic hypofunction impairs postoperative cognition. The preventive effect of taVNS on dNCR may be due to both peripheral cholinergic and central cholinergic mechanisms. For peripheral cholinergic mechanism, based on the current data in this study, we assume that decreased cholinesterase activity alleviates its inhibitory effect on CAIP in peripheral immune cells (macrophages), thereby reducing levels of peripheral inflammatory factors that penetrate the BBB and alter neuronal function. The central cholinergic nervous system is known to be critical for the development of POCD [54] and Alzheimer’s disease [55]. In addition, peripheral BChE activity is considered a reasonable indicator of brain BChE [56]. taVNS could increase ACh levels in brain [53]. Therefore, taVNS may be enhancing the central cholinergic neurotransmission in neuronal cells to prevent dNCR. Notably, we did not examine the change of central cholinergic system, and further research is required to test the prediction of this view.

This study had several limitations. First, there is no widely accepted standard for optimal stimulation parameters for taVNS. In this study, we set parameters based on previous animal experiments [57]. Further studies are necessary to evaluate the shortest time period and the lowest stimulation strength of taVNS that would have a beneficial effect. Second, the cytokines were measured from peripherally obtained blood, which may not necessarily correspond to values in the brain. Third, cytokines were only measured once after surgery. A long observation period may be of importance in determining the relationship of the dNCR with the level of inflammatory factors and the activity of cholinesterase. Finally, the small sample size is a further potential limitation.

Conclusions

The taVNS exerted beneficial effects on dNCR in elderly patients at 1 week after TJA, which may be related to the inhibition of inflammatory cytokine production and the reduction of cholinesterase activity. The novel non-pharmacological neuromodulation approach for preventing dNCR would be welcome.