This study was designed to be a real-world retrospective study to investigate the survival outcomes of neoadjuvant sintilimab combined with chemotherapy in patients with locally advanced ESCC at The Fourth Hospital of Hebei Medical University. The study was approved by the Ethics Committee of our hospital and conducted in accordance with the 2013 edition of the Declaration of Helsinki. Written informed consent were waived. The inclusion criteria were included: patients with histologically confirmed locally advanced resectable ESCC, aged 18 years or older, both sexes, with clinical stage II-IVA, treating with neoadjuvant sintilimab combined with albumin-bound paclitaxel and nedaplatin followed by surgery, completed at least 1-year follow-up. Exclusion criteria included: having other anti-tumor treatments before or during the neoadjuvant treatment. Pretreatment baseline staging was according to American Joint Committee on Cancer (AJCC) 8th edition TNM staging system. Chest-abdominal contrast enhanced computed tomography (CT), esophageal enhanced magnetic resonance imaging (MRI), endoscopic ultrasound (EUS), and cervical ultrasound were performed. Position emission tomography positron emission tomography computed tomography (PET-CT) was also performed when necessary.

All eligible patients completed 2-4 cycles of neoadjuvant treatment with sintilimab (200mg, I.V, D1), albumin-bound paclitaxel (260mg/m (2), I.V, D1) and nedaplatin (80mg/m (2), I.V, D1) of each 3-week cycle. Dose reductions were not permitted for sintilimab. Sintilimab was discontinued if a grade 2 or specific grade 3 treatment-related adverse event (TRAE) occurred during treatment and resumed when the TRAE reduce to grade 1 or below. Sintilimab was permanently discontinued if a specific grade 3 or grade 4 TRAE occurred. Chemotherapy was discontinued when patients experienced a grade 3–4 TRAE and resumed when the TRAE reduce to grade 1 or below. The chemotherapy dose at resuming was reduced to 75% of the initial dose when the grade 3–4 AE occurred for the first time and continued to reduce to 50% of the initial dose when the grade 3–4 AE reoccurred. The chemotherapy was permanently discontinued if the grade 3–4 AE reoccurred despite 2 dose reductions. Radiologic responses and clinical restaging were assessed every two cycles by the same means as the baseline. McKeown esophagectomy with two- or three-field lymphadenectomy was selected according to patients’ condition and performed for all suitable patients. Pathological examination was carried out according to the standard protocols. Pathological response was assessed by tumor regression grade (TRG) referred to the Becker system (33). The adjuvant regimens include albumin-bound paclitaxel and nedaplatin, sintilimab monotherapy, or sintilimab combined with albumin-bound paclitaxel and nedaplatin. The decision was made through the multidisciplinary board according to the original response to immune-chemotherapy and postoperative clinical conditions. The adjuvant regimen was decided by selected by multidisciplinary team according to the efficacy and safety of neoadjuvant treatment and postoperative recovery, including adjuvant albumin-bound paclitaxel and nedaplatin, sintilimab monotherapy, or sintilimab combined with albumin-bound paclitaxel and nedaplatin. pCR patients may not receive adjuvant treatment. Follow-up were routinely conducted every 3 months during first 2 years after surgery, and then every 6 months after 2 years.

The primary outcome was disease-free survival (DFS) at 24 months. DFS was defined as the time from the date of neoadjuvant treatment to recurrence or death by any cause. Disease recurrence included locoregional recurrence (LRR) and distant recurrence (DR) that occurred after R0 resection. Locoregional recurrence was defined as the recurrence within the esophagus, anastomosis or regional lymph nodes. Distant recurrence was defined as distant organ metastases, peritoneal carcinomatosis or recurrence within nonregional lymph node. The secondary outcomes were overall survival (OS) at 24 months. OS was defined as the time from the date of neoadjuvant treatment to death by any cause. Impact of pathological response (pCR, MPR and tumor downstaging) on DFS was exploratory analyzed. pCR (corresponds with Becker TRG 1a) was defined as no evidence of residual tumor cells in the resected primary tumor and axillary lymph nodes. MPR (corresponds with Becker TRG 1a+1b) was defined as 10% or fewer residual tumor cells in the primary tumor. Tumor downstaging was defined as a decrease in T or/and N of pre-surgery stage after neoadjuvant (ypTNM) or post-surgery pathological stage (pTNM) relative to baseline clinical stage (cTNM). Upstaging was defined as an increase in T or/and N of ypTNM or pTNM stage relative to baseline cTNM.

Statistical analysis was performed using SPSS software (IBM SPSS Statistics, RRID: SCR_016479 version 26.0) and R software (RRID: SCR_001905 version 4.0.0). The continuous variables were presented as median and range. The categorical variables were presented as number and percentage. The pCR, MPR, tumor downstaging and R0 resection rate with 95% CI were calculated using the Clopper–Pearson exact method based on binomial distribution. Median follow-up time was calculated using the reverse Kaplan–Meier method. DFS and OS and corresponding 95% CIs were estimated using the Kaplan-Meier method, and a log-rank test was used for comparisons between pathological response subgroups. Univariable and multivariable Cox regression models were used to estimate the effect of neoadjuvant treatment among subgroups according to baseline characteristics. All statistical testing is two-tailed and performed at the 5% significance level.

Ninety eligible patients were included in the analysis between July 2019 and October 2021. The major characteristics of the patients are shown in Table 1 . Sixty-one (67.8%) were male, the median age was 65 years (range 49-78 years), most tumors were located in the middle esophagus (44.4%) and lower esophagus (42.2%), 48 (53.3%) patients had Eastern Cooperative Oncology Group (ECOG) performance status (PS) 0 and 42 (46.7%) had ECOG PS 1 or 2. Thirty-two (35.5%) patients had clinical stage II, 52 (57.8%) patients had clinical stage III, and 6 (6.7%) had clinical stage IVA.

The median number of neoadjuvant cycles was 3 (range 2-4). Forty-two (46.7%) patients received 2 cycles of the neoadjuvant treatment, and another forty-eight (53.3%) received 3–4 cycles. Eighteen patients experienced chemotherapy dose reduction due to TRAEs. Five patients experienced treatment discontinuation due to TRAEs, including one patient because of Immune-mediated colitis, one patient because of liver Abnormalities, two patients because of hyperthyroidism, and two patients because of leukopenia and neutropenia. No treatment-related surgical delay or death was observed.

All 90 patients underwent scheduled surgery, all patients achieved R0 resection. The median interval between the end of neoadjuvant therapy and surgery was 34 days (range 21-95 days). The median number of resected lymph node was 29 (range 20-52). Seventy-five (83.3%) patients completed two field lymphadenectomy, and 15 (16.7%) completed three field lymphadenectomy. The median operation time was 303 minutes (range 142-563 minutes), and the median intraoperative blood loss was 150ml (range 100-2300 ml). The treatment responses are summarized in Table 2 . Postoperative pathological analysis showed that the pCR (TRG1a) rate was 31.1% (95CI 21.8%-41.7%), MPR (TRG1a+1b) rate was 61.1% (95CI 50.3%-71.2%). The median length of hospital stay was 12 days (range 8-54 days).

Sixty (66.7%) patients received adjuvant therapy, including 50 (55.6%) patients receiving sintilimab combined with chemotherapy, 5 (5.6%) patients receiving sintilimab, and 5 (5.6%) patients receiving chemotherapy. The median cycle of adjuvant sintilimab was 8 (range, 1-17), The median cycle of adjuvant chemotherapy was 2 (range, 1-4).

As of data cut-off, the median follow-up was 21.1 months (range 14.0-39.0). 15 (16.7%) patients had disease recurrence and 9 (10.0%) of them died, in addition, 1 (1.1%) patient died of non-cancer-related cause without recurrence. 2 (2.2%) patients experienced locoregional recurrence, 12 (13.3%) patients experienced distant recurrence, and 1 (1.1%) patient experienced concurrent locoregional recurrence and distant recurrence ( Table 3 ). Of the 30 patients who were followed up for more than 24 months, only 3 recurred 15.8 months, 16.9 months and 20.8 months after treatment, respectively.

The median DFS and median OS had not been reached yet, 1-year DFS rate was 91.1% (95%CI 85.4%-97.2%), and 2-year DFS rate was 78.3% (95%CI 68.8%-89.1%). 1-years OS rate was 97.8% (95%CI 94.8%-100.0%), and 2-years OS rate was 88.0% (95%CI 80.6%-96.0%) ( Figures 1 , 2 ). Subgroup analysis showed that postoperative pathological stage, pCR, MPR, tumor pathological down-staging were significantly correlated with survival outcome ( Table 4 , Figure 3 ). Univariable and multivariable Cox regression analysis according to baseline characteristics identified cycle number of neoadjuvant treatment as independent predictor of DFS ( Table 5 ). Patients who completed 3–4 cycles of neoadjuvant treatment increase survival compared to those received 2 cycles, 2-year DFS rate was 88.1% (95%CI 76.9%-100%) and 68.0% (95%CI 53.1%-85.6%) respectively ( Figure 4 ).