This single-center, randomized, double-blind, placebo-controlled trial was conducted at the First Affiliated Hospital of Soochow University, Suzhou, Jiangsu, China. The study protocol was approved by the Ethics Committee of the First Affiliated Hospital of Soochow University (Approval No. 2020-127). This study was registered on the Chinese Clinical Trial Registry on May 9, 2021 before the enrollment of the first participant. All patients provided their written informed consent. This study was performed following the Declaration of Helsinki and the Consolidated Standards of Reporting Trials guidelines.

Patients aged ≥18 years with American Society of Anesthesiologists (ASA) physical status I–III scheduled for thoracoscopic lung cancer surgery were eligible for inclusion. Preoperative diagnosis of lung cancer was based on computed tomography or pathological results. The exclusion criteria included:

Based on a central randomization system,¹ patients were randomly allocated in a 1:1 ratio to either the esketamine group or the normal saline group. The details of allocation were concealed using password protection. A research nurse not involved in the subsequent study prepared the study medications according to the randomization results. Both esketamine and normal saline placebo are clear and colorless fluids, so there is no way to distinguish them. All patients, anesthesiologists, surgeons, perioperative care providers, and postoperative assessors were blinded to the group allocation until the final statistical analysis.

After entering the operating room, patients were monitored with electrocardiography, non-invasive blood pressure, and peripheral oxygen saturation (SpO₂). Under local anesthesia with 2% lidocaine, an arterial line was placed via the radial artery for continuous arterial pressure monitoring and blood sampling. Anesthesia depth was assessed using the bispectral index (BIS). Anesthesia was induced with intravenous propofol 1.5–2 mg/kg, sufentanil 0.3–0.5 μg/kg, and cisatracurium 0.2 mg/kg. Patients were intubated with a double-lumen endotracheal tube, and the location was confirmed by capnography, auscultation and fiberoptic bronchoscopy. In the lateral position, single-lung ventilation was started (tidal volume of 4–6 ml/kg, respiratory rate of 12–20 breaths/min, 60–100% oxygen, and positive end-expiratory pressure of 5–10 cm H₂O) to maintain end-tidal carbon dioxide within 35–45 mmHg and SpO₂ ≥ 90%. Lung recruitment maneuvers were carried out when indicated. Anesthesia was maintained with sevoflurane 1–3% inhalation, titrated to BIS values within 40–60. Intermittent boluses of sufentanil or cisatracurium were given intravenously when needed. Patients received Lactated Ringer’s solution for intraoperative volume repletion. Nasopharyngeal temperature was maintained within 36–37°C using a warming blanket.

Hypotension was defined as mean arterial pressure (MAP) < 65 mmHg and treated with fluid infusion, ephedrine, or phenylephrine. Bradycardia was defined as heart rate (HR) < 50 beats/min and treated with intravenous atropine. Interventions for hypertension (defined as MAP >90 mmHg) and tachycardia (defined as HR > 100 beats/min) included adjustment of anesthesia and administration of medications (nitroglycerine, nicardipine or esmolol). At the end of surgery, all patients received intravenous flurbiprofen axetil 50 mg, ondansetron 4 mg, and wound infiltration with 0.5% ropivacaine 15–20 ml. Patients were transferred to the post-anesthetic care unit and extubated there.

In the esketamine group, esketamine 0.1 mg/kg was administered intravenously at about 3 min before skin incision, followed by a continuous infusion of 0.1 mg/kg/h until the end of surgery. The normal saline group received the same volume of normal saline placebo. Patients received sufentanil-based patient-controlled intravenous analgesia (PCIA) during the first 48 h postoperatively. In the PCIA device, sufentanil 2 μg/kg combined with esketamine 1 mg/kg (the esketamine group) or sufentanil 2 μg/kg only (the normal saline group) was diluted with normal saline to a final volume of 100 ml. The background infusion rate was 1.5 ml/h, the bolus dose was 1.5 ml, and the lockout interval was 10 min. Thus, we used esketamine 0.1 mg/kg bolus +0.1 mg/kg/h intraoperatively, followed by 0.015 mg/kg/h postoperatively for 48 h. This dosing regimen of esketamine was according to our clinical practice and the recent literature in which esketamine was administered as 0.125 to 0.25 mg/kg bolus, with or without 0.015 to 0.125 mg/kg/h for 48 h (11, 19, 20).

The primary outcome was the incidence of depressive symptoms at 1 month postoperatively. An independent investigator assessed the depressive symptoms using a WeChat-based questionnaire of the Beck Depression Inventory-II (BDI-II) (18). The BDI-II is a validated self-report measure of depressive symptoms, consisting of 21 items (3 points for each item and scores ranging from 0–63) (21). A total score of 0–13, 14–19, 20–28, and 29–63 indicates absent, mild, moderate and severe depressive symptoms, respectively (22). Patients completed the questionnaire in their own time to avoid the possibility of observer bias (23).

The secondary outcomes included BDI-II scores and depressive symptoms incidence at 48 h postoperatively, at hospital discharge, and at 3 months postoperatively, BDI-II scores at 1 month postoperatively, Beck Anxiety Inventory (BAI; 21 items with scores ranging from 0–63) (24) scores and anxious symptoms incidence at 48 h postoperatively and at hospital discharge, Quality of Recovery-15 (QoR-15) (25) scores at 48 h postoperatively, at hospital discharge, and at 1 month postoperatively, and postoperative 1- and 3-month mortality. BAI and QoR-15 questionnaires were also completed via the WeChat application. Studies have validated the Chinese versions of BDI, BAI, and QoR-15 (26–28).

Patients’ demographics and baseline characteristics were collected 1 day before surgery. Patients were assessed for preoperative depressive and anxious symptoms using BDI-II and BAI. Intraoperative data included use of anesthetics and analgesics (propofol, sevoflurane, sufentanil and esketamine), hemodynamic events, blood loss, fluid infusion, urine output, type of surgery, duration of surgery, and duration of anesthesia. Postoperative data included pain scores at rest and on coughing at 24 and 48 h postoperatively, patient-controlled sufentanil consumption within 0–24 h and 0–48 h postoperatively, esketamine consumption during 0–48 h postoperatively, adverse events (postoperative nausea and vomiting [PONV], nightmare, dizziness, blurred vision, dissociation and delirium), and length of postoperative hospital stay. Pain intensity was assessed using the visual analog scale (VAS; ranging from 0–10, higher scores indicate more pain). Delirium was assessed using the Confusion Assessment Method (CAM) or Confusion Assessment Method for Intensive Care Unit (CAM-ICU) (29). Patients were followed up at 1 month and 3 months postoperatively via WeChat questionnaires.

A recent study reported that the incidence of postoperative depressive symptoms was 19% in patients undergoing surgical treatment for lung cancer (30). The effect of esketamine on depressive symptoms after lung cancer surgery is unknown. A recent study showed that a low-dose ketamine compared to normal saline placebo reduced the occurrence of depressive symptoms by 61% (from 17.5 to 6.9%) at 30 days after major surgery in older adults (9). In addition, another study suggested that the postoperative depression scores were lower in the esketamine group than those in the ketamine group for patients with breast cancer (11). Based on these, we hypothesized that use of esketamine could reduce the depressive symptoms by 70% (i.e., from 19 to 5.7%) in our patients. Thus, 70 patients in each group were required with one-sided test at α level of 0.05 and power of 80%. To allow for a 10% dropout rate, we finally enrolled 156 patients with 78 in each group. We calculated the required sample size using the PASS software (NCSS, LCC, Kaysville, UT).

Data are expressed as mean [standard deviation (SD)], median [inter-quartile range (IQR)], or numbers (percentages). Continuous data were checked for normality using the Kolmogorov–Smirnov test and analyzed using the unpaired Student t-test or Mann–Whitney rank–sum test. Categorical data were analyzed with the chi-square test or Fisher exact test. For the primary and secondary outcomes, the between-group differences were analyzed using risk difference or median difference with 95% confidence interval (CI).

The primary analyzes were conducted in the modified intention-to-treat population including patients who underwent randomization and surgery with available primary outcome measurement (1-month depressive symptoms rate). As depressive symptoms were associated with the pathological results, the primary outcome was also analyzed after excluding patients without lung cancer diagnosis.

We analyzed the risk factors for depressive symptoms at 1 month postoperatively in the modified intention-to-treat population. First, we applied the univariate logistic regression analysis for possible predictive factors. We chose the variables with p < 0.25 based on statistical significance and clinical relevance, and we included these variables into the multiple regression analysis. We analyzed the variance inflation factor to assess multicollinearity of the variables in the multiple regression model. The goodness of fit of the model was examined using the Hosmer–Lemeshow test.

No multiple testing corrections were planned for the secondary outcomes, so these results should be regarded exploratory. We did not perform an interim analysis. Missing values were not imputed. In addition, we conducted post hoc subgroup analysis for the 1-month depressive symptoms rate between male and female patients. Except for the subgroup analysis, all analyzes in this study were planned a priori. Statistical analyzes were performed with SPSS software (version 22.0, IBM SPSS, Chicago, IL). A two-sided p < 0.05 was considered statistically significant.

From May 18, 2021 to May 30, 2022, a total of 332 patients scheduled for thoracoscopic lung cancer surgery were screened (Figure 1). Of them, 176 patients were excluded due to not meeting the inclusion criteria (n = 70), refusal for participation (n = 101), and cancelation of surgery (n = 5). Thus, 156 patients were randomized, with 78 in each group. Five patients were lost to 30-day follow-up (3 in the esketamine group and 2 in the normal saline group). The modified intention-to-treat analyzes included 151 patients (75 in the esketamine group and 76 in the normal saline group) who had available primary outcome data. Analyzes were also performed after excluding patients without lung cancer diagnosis (3 in the esketamine group and 2 in the normal saline group). At 90 days postoperatively, 74 patients in the esketamine group and 76 in the normal saline group were assessed for depressive symptoms.

The demographic and baseline characteristics were comparable between the two groups (Table 1). The mean (SD) age was 54 (12) years for both groups. 69.2% of patients in the esketamine group and 76.9% in the normal saline group were female sex. At 1 day before surgery, patients were assessed for depressive and anxious symptoms. Two patients in the normal saline group showed preoperative depressive symptoms; however, they had no history of previous depressive episodes, nor they were on antidepressants. Three patients in the esketamine group and 5 patients in the normal saline group exhibited anxious symptoms. The median (IQR) time between suspected diagnosis and surgery was 2.5 (0.6, 7.3) and 2 (0.5, 9.3) months in the esketamine group and the normal saline group, respectively.

Table 2 shows the intraoperative and postoperative data. The two groups had similar intraoperative propofol, sevoflurane, and sufentanil consumption. The mean (SD) total dose of intraoperative esketamine was 21.6 (7.0) mg. The two groups were similar in terms of hemodynamic events, blood loss, fluid infusion, urine output, type of surgery, duration of surgery, or duration of anesthesia. For the postoperative data, the two groups had similar VAS pain scores and required similar amount of patient-controlled sufentanil. The mean (SD) total dose of postoperative esketamine was 45.1 (8.8) mg. Adverse events were comparable between groups. 43.6% of patients in both groups experienced PONV; most episodes were mild, and 3 patients required additional antiemetics. No patients experienced dissociation or delirium. The median (IQR) length of postoperative hospital stay was 4 (3.0, 6.3) days in the esketamine group and 5 (3, 6) days in the normal saline group.

The primary outcome of depressive symptoms at 1 month postoperatively was assessed in two study populations, respectively, (Table 3). In the modified intention-to-treat population, patients receiving esketamine had a significantly lower incidence of depressive symptoms at 1 month postoperatively than those receiving normal saline placebo (1.3% vs. 11.8%; risk difference = −10.5, 95%CI = −19.6% to −0.49%; p = 0.018). Postoperative pathological diagnosis confirmed non-cancerous lesions but not lung cancer in 5 patients (3 in the esketamine group and 2 in the normal saline group). After excluding these patients, the incidence of depressive symptoms was also lower in the esketamine group (1.4% vs. 12.2%; risk difference = −10.8, 95%CI = −20.2% to −0.52%; p = 0.018). The post hoc subgroup analysis showed that the 1-month depressive symptoms rate did not differ significantly between males and females (Figure 2).

The details of patients with depressive symptoms are presented in Supplementary Table S1. These patients aged between 34 and 70 years, and most of them (9 out of 10) were female sex. At 1 month postoperatively, the depressive symptoms were mild in 7 patients, experienced moderate in 2 patients, and severe in 1 patient. Two patients with preoperative depressive symptoms (case 6 and case 9) still had symptoms at 1 month postoperatively. Three patients (cases 1, 3, and 6) needed postoperative psychological counseling and received antidepression medications.

The secondary outcomes are shown in Table 3. The two study groups were similar in terms of the BDI-II scores and incidence of depressive symptoms at 48 h postoperatively and at the time of hospital discharge. Of note, at 1 month postoperatively, the median (IQR) BDI-II score was 5 (4, 7) in the esketamine group, which was similar to 5 (4, 8) in the normal saline group (p = 0.664). At 3 months postoperatively, 1 patient in the normal saline group still had depressive symptoms. Two patients in the normal saline group showed anxious symptoms at 48 h postoperatively and at hospital discharge. Regarding the quality of recovery after surgery, the two groups had similar QoR-15 scores at 48 h postoperatively and at hospital discharge; however, the esketamine group had higher median QoR-15 scores at 1 month postoperatively compared to the normal saline group (median difference = 2, 95%CI = 0 to 5; p = 0.048). One patient in the esketamine group died at postoperative day 45 due to respiratory failure.

Based on the univariate logistic regression analysis, female sex, in paid employment, history of hypertension, and preoperative anxious symptoms were associated with increased risk for depressive symptoms at 1 month postoperatively, while use of esketamine was associated with a lower depressive symptoms rate (Table 4). Thus, these 5 factors were included into the multiple regression model and analyzed as categorical variables.

In this multiple logistic regression analysis, the values of variance inflation factor were < 2 for all factors (1.058 for female sex, 1.126 for in paid employment, 1.167 for history of hypertension, 1.017 for preoperative anxious symptoms, and 1.012 for use of esketamine), which indicates minimal collinearity among these predictors. The Hosmer–Lemeshow test showed that p = 0.698, suggesting a high goodness of fit of our model. The independent risk factors for depressive symptoms at 1 month postoperatively were hypertension (multivariate odds ratio = 6.75, 95%CI = 1.13 to 40.31; p = 0.036) and preoperative anxious symptoms (multivariate odds ratio = 23.83, 95%CI = 3.41 to 166.33; p = 0.001). The use of esketamine was a protective factor against 1-month depressive symptoms postoperatively (multivariate odds ratio = 0.09, 95%CI = 0.01 to 0.91; p = 0.042; Table 5).