Survival impact of perioperative changes in prognostic nutritional index levels after esophagectomy

From January 2009 to May 2019, 188 patients with ESCC at the Department of Surgery, Hamamatsu University School of Medicine, were retrospectively reviewed. All patients underwent esophagogastroduodenoscopy (EGD) and computed tomography (CT) from the neck to the pelvis to determine the clinical stage. The clinical stage was diagnosed based on the Union for International Cancer Control TNM classification of malignant tumors, 8th edition [15].

Patients who met the following criteria were enrolled in this study: (1) age > 20 years, (2) Eastern Cooperative Oncology Group performance status of 0–1, (3) histological diagnosis of ESCC by endoscopic biopsy, (4) no double cancer, (5) radical esophagectomy, (6) survival for at least 90 days after surgery, and (7) patients who survived and were followed up for more than 2 years. Patient ineligible for study enrollment was based on the following exclusion criteria: death within 90 days after surgery (n = 2), double cancer (n = 10), salvage surgery (n = 12), and interruption of follow-up within 2 years (n = 6). Finally, 158 patients were included in the study (Fig. 1).

PNI was calculated as 10 × serum albumin (g/dL) + 0.005 × total lymphocyte counts (per mm³) [12]. Preoperative PNI levels were measured just before the surgery. When patients were administered neoadjuvant chemotherapy, the PNI level was measured after this treatment. Postoperative PNI levels were obtained 1 month after surgery. The time-dependent receiver operating characteristic (ROC) analysis of preoperative PNI-based prediction of 5-year survival revealed that the cutoff value of preoperative PNI was 46.9 (sensitivity = 0.613, specificity = 0.651, Youden index = 0.264) (Online Resource 1). Moreover, the time-dependent ROC analysis of postoperative PNI-based prediction of 5-year survival revealed that the cutoff value of postoperative PNI was 40.9 (sensitivity = 0.572, specificity = 0.540, Youden index = 0.112) (Online Resource 2). Based on these results, patients were divided into preoperative-high group (Group H) and preoperative-low group (Group L). Furthermore, patients in Group L were divided into preoperative-low and postoperative-high group (Group L–H) and preoperative-low and postoperative-low group (Group L–L) (Fig. 1).

Neoadjuvant chemotherapy was performed as a standard treatment for patients with non-stage I ESCC. The treatment regimen was a combination of cisplatin and 5-fluorouracil or a combination of docetaxel, cisplatin, and 5-fluorouracil. Transthoracic esophagectomy with 2- or 3-field lymph node (LN) dissection and gastric conduit reconstruction via the posterior mediastinal route was performed as a standard surgical procedure at our institution [16]. In the thoracic approach, video-assisted thoracoscopic surgery in the left decubitus position was generally adopted. Thoracotomy was performed for patients who refused thoracoscopy or were enrolled in another clinical trial. When the stomach could not be used due to a previous history of gastrectomy, reconstruction using the right hemi-colon was performed [16]. Only one patient in Group L–H underwent reconstruction using the right hemi-colon in a two-stage operation because of a history of gastrectomy due to gastric ulcer and scleroderma. Postoperative complications were evaluated for pneumonia, anastomotic leakage (AL), and surgical site infection (SSI) using the Clavien–Dindo classification. Complications of grade 2 or higher were identified as postoperative complications [17, 18]. The multidisciplinary support team, comprised surgeons, physicians of rehabilitation, nurses, dentists, physiotherapists, speech–language–hearing therapists, and managerial dieticians. This team was set up since April 2017, to prevent complications and ameliorate the postoperative nutritional status of all patients after esophagectomy [17].

According to our protocol, patients were instructed to abstain from smoking and alcohol consumption for at least 4 weeks before surgery. Physiotherapists commenced respiratory rehabilitation using an incentive spirometer. Following the surgery, early ambulation and restart of respiratory rehabilitation were encouraged. An elemental diet was started at 10 kcal/h from the day of surgery via a jejunostomy tube for patients with a jejunostomy. The tube-feeding dose was gradually increased to 1200 kcal/day. Enteral nutrition through jejunostomy was continued in the patients, even after the hospital discharge until there were satisfactory results in oral intake [17].

At the time of oral feeding initiation, the rehabilitation physician and the speech–language–hearing therapists performed videofluoroscopic (VF) and videoendoscopic (VE) examinations, especially for the patients receiving multidisciplinary team support. The meals were started with a dysphagia diet. The diet was changed to liquids in accordance with the improvement in the swallowing function. Furthermore, the total calorie intakes were based on decisions made by the multidisciplinary team conference. The calorie intake at the time of discharge was approximately 1500 kcal/day after introduction of multidisciplinary team support [17]. After the first visit in the outpatient clinic, the dietician evaluated the patients’ dietary intake and continued to give nutritional advice regularly.

Postoperative follow-up was performed using CT every 6 months and EGD every year for 5 years after surgery. Recurrence-free survival (RFS) was calculated from the operation to the day of recurrence of esophageal cancer. Overall survival (OS) was calculated from the operation to the day of death due to esophageal cancer. Patients were followed up until death or until the end of the study (May 31, 2021). Patients who died of an illness unrelated to esophageal cancer, interrupted follow-up, or under following up were recognized as censored, and RFS and OS were calculated based on the days until censoring.

The time-dependent ROC analysis was performed using R software v 4.0.5 (R Foundation for Statistical Computing, Vienna, Austria [19]). The time-dependent ROC analysis had used to diagnosis 5-year survival in this study. Other statistical analyses were performed using IBM SPSS Statistics version 26 for Windows (Chicago, IL, USA). Medians and ranges were calculated, and differences were identified using Student’s t test. The Mann–Whitney U test was used for non-parametric analyses. Differences between each category were identified using the Chi-square test or Fisher’s exact test, and the Kruskal–Wallis test was used to identify the differences in continuous variables. Survival curves were produced using the Kaplan–Meier survival method and log-rank test. Hazard ratios (HRs) were calculated, and univariate and multivariate analyses were performed using Cox proportional hazards regression models. The threshold for significance was set up at p < 0.05.

Clinicopathological features between Group H and Group L are shown in Table 1. The median follow-up period was 43.4 (4.3–133.5) months. Patients in Group L had a more advanced clinical stage (p < 0.001) and, thus, the rate of patients who received neoadjuvant chemotherapy was higher in Group L (p = 0.004). The rate of multidisciplinary team support was similar in both groups (p = 0.736) (Table 1). Radical esophagectomy with two- or three-field LN dissection was performed in all patients. There was no significant difference in the rate of minimally invasive esophagectomy (p = 0.338). Moreover, no significant difference was observed in terms of LN dissection, operation time, and jejunostomy (Table 1). The rate of reconstruction organ was similar among both groups. With postoperative complications, the prevalence of postoperative pneumonia was significantly higher in Group L than that in Group H (39.1% vs. 23.6%, p = 0.038). There was no significant difference in AL and SSI between Groups L and H (AL 20.3% vs. 21.3%, p = 1.000; SSI 29.0% vs. 30.3%, p = 1.000). Pathological study showed that the patients in Group L were diagnosed at an advanced stage (p = 0.002). The rate of patients who received either adjuvant chemotherapy or radiation was higher in Group L than those in Group H (p = 0.012). There was no significant difference in the proportion of local recurrence between the groups. However, the rates of patients who had regional LN and distant organ recurrence were significantly higher in Group L than in Group H (regional LN recurrence: 31.9% vs 17.0%, p = 0.037; distant organ recurrence: 34.8% vs 17.0%, p = 0.015, respectively) (Table 1).

In Kaplan–Meier analysis, the OS and RFS were significantly worse in Group L than in Group H (p = 0.001, p = 0.003) (Fig. 2).

To confirm the correlation between the changes in PNI level from the preoperative to postoperative phase and survival, the survival rates among the three groups were compared. Clinicopathological features between Groups L–H and L–L are shown in Table 2. There was no significant difference in clinical stage between Groups L–H and L–L (p = 0.132). The proportion of multidisciplinary team support in Group L-L tended to be lower than that in Group L–H (25.6% vs. 40.0%, p = 0.298). The incidence of postoperative infectious complications (AL, pneumonia, and SSI) was higher in Group L–L than in Group L–H (p = 0.032). The prevalence of AL and SSI was significantly higher in Group L–L than those in Group L–H (AL: 30.8% vs. 6.7%, p = 0.016; SSI: 41.0% vs. 13.3%, p = 0.016). The rate of patients who received enteral nutrition in the early postoperative phase was similar between the groups (53.3% in Group L–H vs. 51.3% in Group L–L, p = 1.000). There was no difference in the population of patients with death unrelated to esophageal cancer (p = 1.000). The proportion of patients who received adjuvant chemotherapy was similar between the two groups (p = 0.336). However, adjuvant therapy was suspended in two patients from Group L–L due to adverse events (Table 2).

In the Kaplan–Meier analysis, the RFS in Group L–H tended to be better than those in Group L–L (p = 0.077). Furthermore, patients in Group L–H had significantly better OS than those in Group L–L (p = 0.023) (Fig. 3). Patients in Group L–L had significantly worse OS and RFS than those in Group H (p < 0.001, p < 0.001). However, there was no significant difference in OS and RFS between Groups L–H and H (p = 0.224, p = 0.161) (Fig. 3).

All of the variables shown in Table 3 were potentially significant clinicopathologic variables of prognosis in patients with ESCC. In univariate analysis, neoadjuvant chemotherapy, advanced pathological stage (≥ II), and Group L–L were associated with a worse OS (p = 0.004, p < 0.001, and p < 0.001, respectively) (Table 3). Postoperative complications (AL, pneumonia, and SSI) were not associated with OS (p = 0.548) (Table 3). In multivariate analysis, advanced pathological stage (HR 10.947; 95% CI 2.590–46.268; p = 0.001) and Group L-L (HR 2.171; 95% CI 1.249–3.775; p = 0.006) were independent predictors of worse OS (Table 3).

A stratified analysis before and after introduction of multidisciplinary team support was performed. The proportion of patients in Group L–H increased markedly from 17.1% before initiation of the multidisciplinary team support to 22.6% after initiation of the multidisciplinary team support (Online Resource 3). Before introduction of multidisciplinary team support, patients in Group L–H showed significantly better OS and RFS than those in Group L–L (median survival time (MST); OS: 61.1 months vs. 35.4 months; RFS: 61.1 months vs. 24.4 months). Furthermore, there was no difference in the OS and RFS between Groups L–H and H. Although the number of patients who received multidisciplinary team support was small, after it was initiated, patients in Group L–H also tended to have a good OS than those in Group L–L (MST of OS: 36.2 months vs. 23.7 months). Moreover, the survival curves were also similar between Groups L–H and H (Online Resource 3).

To investigate the survival impact postoperative nutritional support, the long-term outcomes of patients with postoperative infectious complications (AL, pneumonia, and SSI) were assessed using a stratified analysis which patients were divided into groups based on whether they developed postoperative infectious complications (Online Resource 4). In the stratified analysis, either with or without postoperative infectious complications, patients in Group L–H were tended to have better OS and RFS than those in Group L–L (Online Resource 4).