Introduction

Recent reports and studies suggest that anesthetic management for cancer surgery might have an impact on the recurrence of neoplasms, metastasis spreading, and the associated long-term outcome [1]. The underlying multifactorial cellular and molecular mechanisms are not yet entirely understood. The current theory regarding the impact of analgesic and anesthetic drugs on cancer biology suggests a combined effect of opioid-related reduced immune response and consecutive inadequate surgical stress response [1]. The transient immune deficiency caused by opioid analgesic and anesthetic drugs [2, 3] or their direct impact on tumor cells may lead to conducive conditions for tumor cell survival in the perioperative period [4, 5]. A treatment free window in the immediate postoperative period where chemotherapy, radiotherapy, or even both are avoided might give micro-metastasis the opportunity to grow, establish, or even further spread early after surgery [6, 7, 5]. The perioperative opioid administration seems to play a key role in the postoperative tumor cell survival, spreading of metastasis, and recurrence of cancer. Opioids inhibit natural killer cell cytotoxicity and impair cellular and humeral immune function in humans [8–10]. The perioperative use of regional anesthesia might attenuate the surgical stress response and consecutively reduce the overall dose of opioids and volatile anesthetics. This might be beneficial in patients undergoing cancer surgery. In melanoma and breast cancer surgery, the additional administration of regional anesthesia leads to a decreased rate of tumor recurrence and a longer cancer-free postoperative interval in retrospective settings [11, 12]. As esophageal cancer surgery is an intervention demanding high doses of postoperative opioids but can also be managed with epidural analgesia, we propose esophageal cancer as an appropriate cohort to investigate the impact of epidural anesthesia. The long-term benefits of epidural analgesia including more than a 5-years observation of oncological outcome in patients undergoing esophageal cancer surgery have never been analyzed before. The immediate postoperative clinical benefits of epidural anesthesia including reduced ICU hospitalization, earlier mobilization, reduced pulmonary complications, and better analgesia were reported for many different surgical cohorts [13–15]. However, a decided analysis of immediate postoperative benefits from epidural analgesia with patients undergoing esophagus cancer surgery is still missing.

Therefore, the aim of this study was to analyze whether patients undergoing esophageal cancer surgery took any immediate postoperative clinical benefit or long-term oncological advantage of postoperative epidural analgesia.

Methods

Due to the retrospective and anonymous character of this study, the local ethics committee disclaimed the need for obtaining approval ahead of the analysis. To achieve an observation period of at least 5 years after surgery, we retrospectively analyzed the records of patients undergoing esophageal cancer surgery between 1995 and 2005. The analysis included pre- and postoperative tumor status, adjuvant therapies, cancer recurrence, and the entire anesthesia and postoperative intensive care unit (ICU) as well as normal ward records. All available data from different sources were anonymized and transferred into an SPSS® (IBM, Armonk, USA) data sheet. All tumor-related data were retrieved from the tumor registry of the Department of Surgery. The anesthesia data was transferred from electronic patient documentation system (PDMS) and paper records to the SPSS® study data base. ICU and normal ward documentation were screened and relevant information was entered from paper records to the study data base.

All operations were performed as abdomino-right-thoracic esophagus resection using the two cavity surgery technique. Other techniques like trans-hiatal resection were not used in our cohort.

The institutional standard operating procedures (SOP) for patients scheduled for esophageal cancer surgery demand the preoperative administration of an epidural catheter in the absence of specific contraindications like refused informed consent by the patient. After placing of the epidural catheter, a test dose of 3 ml bupivacaine 0.25 % was administered. Anesthesia was performed as balanced anesthesia using end-tidal sevoflurane concentration of 1.5–2.0 % and repetitive injections of fentanyl or continuous infusion of remifentanil. Injection of up to 9 ml ropivacaine 0.375 % and repetitive boluses of 3 ml were administered at the end of surgery. Paracetamol or metamizol were given to every patient at the end of surgery. After finishing surgery, patients were transferred to the ICU and extubated as soon as possible. Postoperative management of epidural catheter anesthesia (PCEA) or, if not, obtained patient-controlled intravenous analgesia (PCIA) was performed by the institutional acute pain team. In the postoperative period, the epidural analgesia was obtained with 0.2 % ropivacaine and, in the absence of specific contraindications, 1 μ sufentanil per milliliter. Oral medication was started when authorized by the surgeon but not within the first 3 days. In both groups, non-opioid analgesic drugs were given as adjuvant analgesic drugs according to the SOP for postoperative pain management. The postoperative opioid consumption of every single drug was retrieved from the records and transferred into the study data base. To achieve a good comparability of the postoperative opioid consumption, all postoperative opioid medication was transcribed to cumulative daily morphine equivalent doses separated as intravenous (i.v.), per os (p.o.), and overall opioid medication [16, 17].

Statistics

Statistical processing is performed as eligible. Gaussian distribution of continuous variables was tested with Kolmogorv-Smirnov test and compared with Student t test if Gaussian distribution is given. Non-Gaussian-distributed data were compared using the Mann-Whitney U test. Categorical variables were given as absolute number and percentage of their occurrence and were compared using the chi square test. Continuous variables are given as median (25 % quartile, 75 % quartile); categorical variables are given as absolute number of occurrence and (percentage). Statistical difference was defined as p < 0.05. Logistic regression and Cox analysis were used as multivariate models.

Results

Overall, this retrospective analysis includes the datasets of 153 patients. Epidural analgesia was administered in 118 patients (77 %). The median duration of epidural analgesia was 6 days (25 % quartile 3 days; 75 % quartile 7 days). In 35 patients (23 %), epidural analgesia was not obtained due to refused consent of the patients or specific contraindications. The main results of the two groups are shown in Table 1. None of the variables addressing preoperative patient or tumor status revealed any statistically significant difference between the group of patients with and the group without epidural analgesia. The duration of surgery given as time from first skin cut to last suture revealed also no statistically significant difference between the two groups.

Table 1 Perioperative and demographic data of the two study groups. Continuous variables are given as median (25 % quartile, 75 % quartile), categorical variables are given as absolute number of occurrence and (percentage)
Full size table

The pathological findings in the resected tumors of the two groups are comparable. R0 resection, indicating the absence of residual tumor was found in 91 % in the non-epidural group and 92 % in group of patients receiving epidural analgesia. The findings for R1 (microscopic residual tumor), R2 (macroscopic residual tumor), and Rx (not assessable) show also no statistically significant difference. The overall postoperative mechanical ventilation hours show significantly longer ventilation duration in the non-epidural group (median 9.4 (25 % quartile 5.1, 75 % quartile 51) h) compared with the epidural group (8.2 (3.9, 17.0) h, p = 0.046).

Regarding the duration of ICU hospitalization, we found a statistically significantly increased ICU duration in the non-epidural group (10 (7, 13) vs. 6 (5, 7), p < 0.001). A multivariate analysis (Table 2) revealed the absence of epidural analgesia as independent (p = 0.04, OR 0.35) risk factor for an extended ICU stay. Extended ICU stay was defined as 75 % quartile of the overall ICU duration (9 days). The analysis included epidural analgesia, age, ASA score 3 or 4, gender, preoperative radiotherapy, and anemia given as hemoglobin below 12 g/dl in female and 12.9 g/dl in male patients. Figure 1 shows the Kaplan-Meier chart of ICU discharge of both, the epidural and the non-epidural group. The presence of an epidural catheter led to a significantly lower opioid consumption in the postoperative course. The cumulative 10-day postoperative i.v. opioid dose in the non-epidural group was 187 mg (90, 543) morphine equivalent compared with 104 mg (6; 219) in the epidural group (p < 0.001). The p.o. opioid consumption was also significantly increased in the non-epidural group compared with the epidural analgesia group (187 mg (72, 2888) vs. 165 mg (80, 275); p = 0.008). A detailed analysis of the influence of opioid consumption on the median survival is listed in Table 3. Log Rank Mantel Cox test showed statistically significant differences between the 0–25 % and the 25–50 % quartile groups (p = 0.025) as well as for the 25–50 % and the 50–75 % quartile groups (p = 0.002).

Table 2 Multivariate analysis addressing the risk of extended (9 days or more) ICU stay
Full size table
Fig. 1
figure 1

Kaplan-Meier chart showing cumulative ICU discharge of patients with and without epidural catheter

Full size image
Table 3 Median survival contingent on the overall opioid consumption
Full size table

Postoperative results addressing the oncological outcome after esophageal cancer surgery showed no statistically significant difference in any of the obtained variables. Cancer recurrence, tumor spreading, and short- and long-term survival show nearly identical results. The cumulative 5-year survival of both groups is shown in the Kaplan-Meier chart of Fig. 2. Preoperative anemia given as hemoglobin below 12 g/dl in female and 12.9 g/dl in male patients was the only clinical parameter which influenced significantly the mortality risk in a multivariate Cox analysis (Table 4). Local tumor relapse and distant metastasis as well as the conjoint combination of both impair the survival significantly.

Fig. 2
figure 2

Kaplan-Meier 5-year survival chart for patients with and without epidural catheter

Full size image
Table 4 Multivariate Cox analysis addressing the risk of mortality related to different clinical and oncological factors
Full size table

Discussion

Experimental studies and clinical reports suggest that epidural analgesia and consecutive reduced opioid requirements might lead to an improved oncological result in esophageal cancer surgery [1, 18]. In the present study, we addressed this question with a retrospective analysis of two patient groups who received abdomino-right thoracic esophagus resection either with postoperative epidural analgesia or without postoperative epidural analgesia.

The demographic data of the two patient groups show no statistically significant difference in any of the preoperative variables as well as preoperative and pathological screened tumor status. This indicates a good comparability and a balanced risk between the two groups. A strong male dominance in the esophagus cancer cohort is often reported by other studies and also found in the present cohort; however, the underlying reasons remain widely unknown. Some differences in substance behavior or hormonal influences are discussed [19].

Early postoperative clinical effects

The results addressing the immediate postoperative course after esophageal cancer surgery demonstrate that patients who received epidural analgesia showed significantly lower intravenous, oral, and overall opioid consumption compared with the non-epidural group. The analysis of the median survival dependent on the opioid consumption showed an association between high overall opioid consumption and a shorter survival. However, in the group with the lowest opioid consumption (0–25 % quartile), the survival is shorter than in the 25–50 % and the 50–75 % quartile groups. Causative for this fact might be that patients who died early after surgery due to postoperative complications received smaller cumulative opioid doses than those who were released alive.

Furthermore, the results reveal a significantly longer duration for mechanical ventilation as well as a significantly increased ICU hospitalization. The lower rate of re-intubation in the epidural patient groups lacks statistical significance but indicates a clear trend to more frequent re-intubations in the non-epidural group. Moreover, patients without epidural analgesia required more frequently and longer postoperative antibiotic therapy. Postoperative re-interventions like thoracic drains and computertomografic-guided target drains were more frequently needed in the non-epidural group. Despite these results’ lack of statistical significance, they might indicate by trend that postoperative epidural analgesia reduces the risk of postoperative complications like pulmonary infections. The early beneficial effects of postoperative epidural analgesia are underlined by results of the multivariate analysis addressing ICU discharges. The absence of epidural analgesia was identified as a significant risk factor for extended ICU stay. These results are representing the well-known effects of epidural analgesia and are in line with former studies addressing the benefits of epidural analgesia in general and especially in patients with esophageal surgery [20–23]. In a 175-patient cohort retrospective analysis, the absence of epidural analgesia for abdomino-thoracic esophageal resection led to a higher rate of postoperative pneumonia, a more frequent rate of re-intubation, a longer ICU stay, and a higher in-hospital mortality in [23]. A reduced ICU hospitalization might not only result in lower costs and lower nosocomial infection but also in a reduced in-hospital mortality [24].

Long-term outcome effects

There are some experimental studies that addressed effects of opioids on cancer recurrence and postulated a positive correlation between opioids and growth of cancer cells [1]. Gong and co-workers found a reliable evidence that opioids have a noticeable effect on regulatory T cells by affecting the function of immune cells or cytokines like tumor growth factor β or interleukin 2 [25]. Other authors address the influence of a micro-opioid receptor (MOR) on angiogenesis and oncogenic signaling and support the hypothesis that opioids affect tumor progression; furthermore, they suggest the MOR as a chemotherapeutic target molecule [26]. Taking these results into account, one might think that a reduced intravenous opioid consumption due to epidural analgesia might lead to an improved oncological outcome. Although our results showed a statistically significant and clinically relevant reduced consumption for orally, intravenously, and over all administered opioids, we found no difference in any oncological outcome variable. Multivariate Cox analysis showed no significant influence of epidural analgesia on mortality. In a retrospective analysis of 140 patients undergoing gastro-esophageal cancer surgery, Hiller and co-workers found no statistically significant effect from intra- and postoperative epidural analgesia on the prevention of cancer recurrence or on 2-year survival [18]. However, an analysis of the Hiller cohort focusing on patients with solitary esophageal cancer and exclusion of those suffering from gastro-esophageal cancer reveals significant recurrence and survival benefits in the 2-year follow-up [18]. In a further trial enclosing 446 patients with prospectively randomized allocation of epidural analgesia for heterogeneous abdominal cancer surgery, Myles and co-authors also found no differences in cancer relapse rates in a long observation period of 9 to 15 years in the overall cohort [27]. A subgroup analysis of 57 patients with esophageal cancer contributing to the Myles cohort showed no significant beneficial effect of epidural analgesia on the 5-year survival (risk ratio 1.02 (0.64–1.64)) [27]. The reported results in other tumor entities were ambiguous: a retrospective analysis of 669 patients undergoing colorectal cancer surgery found recurrence and survival benefits from epidural analgesia only for patients over 64 years, the overall results showed no difference [28]. Holler and co-workers found a substantial benefit of epidural analgesia in colorectal cancer surgery. This benefit was even greater in ASA 3 or 4 patients [29]. In a cohort with over 4000 patients undergoing prostatectomy for cancer treatment, Roiss and co-workers found no beneficial effects of regional anesthesia on the survival [30]. In synopsis, our study, as well as other studies, was unable to demonstrate a clear benefit of epidural analgesia on oncological outcome variables like tumor spreading, cancer recurrence, and survival. Whether the location of cancer (colorectal vs. esophageal vs. prostate) influences long-term effects of epidural analgesia remains unclear.

It is a remarkable secondary result of the multivariate Cox analysis that preoperative anemia is associated with poorer survival. However, it is well known that preoperative anemia and sequent allogenic red cell transfusions are associated with an increased risk of mortality [31].

Limitations

One has to consider that the diversity of influences on postoperative oncological outcome variables and the multiple internal and external influences on intensive care treatment which are not fully covered by the statistical processing might have led to underfitting of the multivariate models and consecutive poor prediction.

In contrary to other reports like the Hiller study, epidural analgesic drugs were only administered in the postoperative period in our cohort. Intraoperative opioids were part of the anesthetic regime in both of our groups; this might reduce the comparability of our results with those of other reports. It might also have seriously reduced potential beneficial effects of epidural analgesia in our cohort. In ovarian cancer patients, de Oliveira found a longer cancer-free survival after primary surgery with intraoperative epidural analgesia but not postoperative neuroaxial analgesia [32]. This report suggests that there might have been different results if epidural anesthesia would have been started during operation and not at the end of the operation. A reliable analysis of the bowel function in our cohort is not suitable as the use of clysters in the postoperative period was individually adapted to the patients and not standardized.

The duration of epidural analgesia differs between the patients in our cohort. Due to the retrospective character of this analysis, there was no study protocol which might have led to a more standardized approach for finishing epidural analgesia. The reasons why establishing of epidural catheter was denied in some patients were not reproducible in every single case. However, the most common reason might have been denied consent of the patients, failed epidural puncture, anticoagulation therapy, or reduced thrombocyte count. A recently published study reports of a good correlation of low preoperative thrombocyte count with impaired long-term survival [33]. These results suggest that the denial of epidural catheter due to reduced thrombocytes might have biased the results of the long-term survival in our study.

There is no information that different opioids lead to severely different effects in cancer biology. Therefore, we pooled the opioid consumption in daily morphine equivalent doses in our study. However, this approach makes a differentiated analysis of the different opioids and their application forms impossible. Epidural-administered sufentanil is known to result in only low systemic plasma levels [34]. However, even this small plasma levels could have affected the oncologic biology in a way that is not yet known.

Conclusions

The results of the present study show that epidural analgesia for esophageal cancer resection leads to a significantly reduced opioid requirement. This finding is consistent for intravenous and oral medication. However, this reduced opioid consumption does not lead to any significant change in the short- or long-term oncological outcome like cancer recurrence, tumor spreading, or overall survival. Nonetheless, epidural analgesia led to the well-known significant advantages in the immediate postoperative outcome as it resulted in decreased mechanical ventilation hours and shorter ICU hospitalization as well as a reduced re-intubation rate. To assess the full impact of epidural analgesia on the oncological outcome after surgery, high-volume register trials with an adequate observation period as well as reliable biochemical models are required. A randomized trial which goes along with sparing epidural analgesia and the deficiency of well-known early postoperative advantages as well as subsequently reduced quality of analgesic therapy for a large number of patients would be ethically not acceptable.