The study protocol was approved by the Ethics Committee of People’s Hospital of Zhengzhou University (Henan, China). The written information consent was provided by each patient before the study. A total of 160 patients having undergone open radical resection of thoracic oesophageal cancer by left thoracotomy were recruited from the People’s Hospital of Zhengzhou University from January 2016 to January 2017. After patients developing hematopoietic dysfunctions, autoimmune diseases, immunodeficiency diseases, severe coronary heart disease, and/or hypertension were excluded, 120 patients were finally recruited for the study. They were randomized via a computer-generated number sequence to receive total intravenous anaesthesia alone (GA group, n = 40), epidural anaesthesia combined with general anaesthesia (EG group; n = 40) or paravertebral blocks based on general anaesthesia (PG group; n = 40). Those patients recruited for the study had no history of endocrine disease and had not undergone radiotherapy, chemotherapy and hormonotherapy before surgery.
Description Of Anaesthesia
Patients were fasted for 6 h from solid food and 2 h from clear liquids. The EG group underwent epidural puncture between T6 and T7 in the left lateral position; subsequently, an epidural catheter was inserted using the paramedian approach and the loss-of-resistance method. A test injection of 3 mL of 2% lidocaine was administered through the epidural catheter when no blood or cerebrospinal fluid was aspirated. Next, 10 ml of 0.375% ropivacaine (Astrazeneca, Wilmington, DE, USA) was injected after the general anaesthesia induction was conducted. For the PG group, the paravertebral block was performed in a left lateral position under ultrasound guidance with a high-frequency linear probe (7–13 MHz) (P07576 type, Sonosite company, Seattle, America). After the back skin was produced using the 2% chlorhexidine solution, the ultrasound probe was covered with a sterile transparent dressing (3M Health Care, St. Paul, Minnesota), and the sterile gel acted as a coupling medium.
The ultrasound probe was placed parasagittal to the long axis of the thoracic vertebrae that was approximately 3 to 4 cm lateral to the spine. When the ultrasound probe was moved, the tip of the transverse process was obviously observed. Subsequently, the paravertebral space was visualized between the parietal pleura and the costotransversarium ligamentum. The in-plane technique was adopted to insert a 16-G needle (Pajunk; Medizintechnologie GmbH, Geisingen, Germany) until the tip of the needle could be observed on the screen. The hydrolocation technique with 0.5 mL normal saline was applied to identify the needle tip until the pleural drift was observed, demonstrating that the needle tip was located in the paravertebral space, and 15 ml of 0.375% ropivacaine (100 mg/10 ml) was injected into the thoracic paravertebral space at T4–5 and T6–7.
Once the level of the regional block reached the expected level for the surgical procedure, general anaesthesia would be induced. All patients in the three groups underwent routine intravenous inductions with 0.2 mg/kg of midazolam (Jiangsu Enhua Pharmaceuticals, Xuzhou, China), 0.05 µg/kg of sufentanil (Hubei Yichang Human-well Pharmaceuticals, Hubei, China) and 0.1 mg/kg of vecuronium bromide (Zhejiang Xianju Pharmaceuticals, Zhejiang, China). Afterwards, the patients were intubated with double-lumen endotracheal tubes, and their lungs were mechanically ventilated.
Anaesthesia was maintained with intravenous infusions of propofol by Target-Controlled Infusion (TCI) and remifentanil, together with vecuronium bromide at 0.1 µg/kg/min (intravenous (i.v.)) when necessary to keep the bispectral index (BIS) at 45 ± 5. The induction and maintenance of anaesthesia for patients in the GA group were identical to those adopted for patients in the EG and PG groups, except for the absence of epidural blocks or paravertebral block.
Monitored Indicators And Experimental Procedures
Venous blood samples (6 mL from each patient) were taken via an intravenous catheter at 30 min before the anaesthesia induction, at the end of the surgical procedure, on the postoperative day POD 1 and on POD 2. A three-channel flow cytometer (FC500; Beckman Coulter, Hialeah, America) was adopted to analyse the survival of T-cell subsets (a cluster of differentiation CD3+, CD4+, CD8+, CD4+/CD8+). The numbers of T-cell subsets were determined with peridinin chlorophyll protein complex-labelled mouse anti-human CD3, fluorescein isothiocyanate-labelled mouse anti-human CD4, as well as PE-labelled mouse anti-human CD8 monoclonal antibodies (Becton Dickinson, Franklin Lakes, NJ, USA). A double antibody sandwich enzyme-linked immunosorbent assay (ELISA) (R and D Systems, Minneapolis, MN, USA) was performed to determine the levels of Cortiso (Cor), (interleukin-6) IL-6, IL-4, tumour necrosis factor-α (TNF-α) and interferon-γ (IFN-γ). The incidences of postoperative nausea and vomiting (PONV), bucking, dizziness, and chest tightness were recorded for 48 h after the surgery. Tropisetron was employed for the symptomatic treatment if vomiting occurred, and the relevant dose was recorded. The visual analog scale (VAS) and the Riker Sedation-Agitation Scale (SAS) were used to assess patients at 20 min after the extubation, at 6 h postoperatively, and on POD 1 and on POD 2. At the end of the operation, the amounts of remifentanil, propofol, and the vasoactive drugs were recorded.
All patients received a 3-year follow-up consisting of a re-examination and a telephone call every 3 months. The local recurrence and distant metastasis of the cancer were recorded, as well as 1-, 2- and 3-year survival rates. The follow-up ranged from the time of diagnosis to January 1, 2019. No patients were lost to follow-up.
The patients were recruited based on CD4 + cells and CD3 + cells in a comparison between PVB and GA published previously  at β = 0.2 (i.e., 80% power). The sample size in each group at α = 0.05 was 35 patients per group (StatMate version 2.00; GraphPad, San Diego, California, USA). Statistical data were analysed with SPSS version 21.0 software (IBM Corp., Armonk, NY, USA), and the data are denoted as the mean ± Standard Deviation (SD). Pairwise comparisons were verified by an independent ttest. If normally distributed, the data were statistically analysed by the analysis of variance, followed by Norman Keul’s test for in-depth comparison; otherwise, the data were analysed by Mann–Whitney test. The survival rate was calculated using KaplanMeier survival curve and a significance analysis of survival rate by a log-rank test. A P value less than 0.05 was considered showing statistical significance.