AL is a severe complication following esophagectomy with high mortality. In spite of increasing research efforts and knowledge gain, pathophysiology and causal factors of AL remains unclear. Notably, the identification of risk factors for AL is pivotal for implementation of patient-tailored treatment regimen. For the intrinsic anatomic reason, lack of esophageal serosa is considered to contribute to the development of AL [12], meanwhile, malnutrition, neoadjuvant therapy and surgical technique may configure as extrinsic factors for the occurrence of AL [13]. Serum albumin is considered as one of the indicators of nutrition and associated with postoperative tissue healing has achieved wide acceptance [14]. In this study, the results suggested that the occurrence of AL is significantly higher in patients with greater perioperative decreased albumin level after treatment with NCT followed by MIE. However, the survival rate was not significant difference between groups, it is indicated that AL does not impair the survival benefit once recovered.

Several attempts had been made to establish a commonly accepted system to predict occurrence of AL, however, no consensus has been reached due to the broad and diversified criterion. As mentioned above, independently from intrinsic anatomic parameter, malnutrition, neoadjuvant therapy and surgical technique are imperative factors in the development of AL [13]. This relates as much to biology of the esophagus as to the surgical technique and perioperative management. To eliminate the heterogeneity of the participants, only ESCC patients treated with NCT followed by MIE were included in this study. Although MIE was demonstrated to be not associated with the occurrence of AL compared to open surgery [15], in this study MIE was performed by the experienced surgeons that can neglect the proficiency gain curve-associated morbidity [16]. Additionally, only patients treated with NCT were included in this study to keep homogeneity of the data.

More recently, management strategies have evolved from prioritizing the intervention after AL occurrence to primarily considering the prevention before AL development. Remarkable progression had been achieved on prediction of AL, e.g., intraoperative real-time monitoring of gastric conduit perfusion which is one of major causes for the development of AL [17, 18]. However, advanced medical devices required is limited implementation only in the high-volume center. Comparing with advanced technique-required methods, detection of serum albumin is a quick and simple way that is routinely handled in the clinical practice.

Hypoalbuminaemia, reflecting a poor nutritional status, which was associated with negative impact on the recovery and survival of patients in different clinical settings. Meanwhile, serum albumin was identified as an excellent parameter in the nutrition assessment and widely used in predicting the prognosis of patients, especially in the postoperative course of surgery. Previously, a certain value of serum albumin (e.g. 35 g/L) was identified as a cutoff for predicting AL after surgery in esophageal cancer patients [19, 20]. However, as a disease highly related to the dietary habits, chronic malnutrition with a low value of serum albumin can be observed in some of esophageal cancer patients, even before diagnosis. Likewise, neoadjuvant therapy is another cause that may lead patients into an unsteady physiological situation, resulting in a malnourished presentation. Furthermore, nutrition support was applied in some of patient with increasing serum albumin level intent preoperatively. However, the value of serum albumin cannot maintain for a long time due to the extrinsic supporting (the half-life of the albumin is about 20 days, and the changes in albumin can occur rapidly, especially in dysfunction of albumin metabolism). Finally, the value of serum albumin may be underestimated by saline infusion postoperatively [21]. For these reasons, serum albumin should be considered as a marker for predicting AL in a broad sense, rather than an instantaneous certain value. Hence, to address the question of whether an instantaneous certain value of serum albumin is accurate enough to predict AL, we investigate the relationship between AL and serum albumin by using the perioperative decreased level of it, which can be most likely to reflect the intrinsic physiological derangement of the patient.

Previous studies use less than 35 g/L, either preoperatively [19] or postoperatively [20], as a threshold value for the serum albumin level associated with the occurrence of AL was determined to provide guidance in the management of patients. However, in current study, nearly 10% (13/137, 9.5%) of patients receiving NCT had a preoperative serum albumin level of less than 35 g/l, meanwhile, almost all patients (135/137, 98.6%) with a lower level than 35 g/l postoperatively. Moreover, no significant difference was exhibited when 35 g/l of serum albumin level was defined as cutoff between groups, irrespective of performing surgery or not (preoperatively, 1/12 vs. 7/117, p = 0.556679; postoperatively, 1/1 vs. 7/128, p = 0.113783). To address the question of why the correlation disappeared in our study and whether this is a surrogate for the effect of neoadjuvant therapy, we chose to investigate this unusual phenomenon by using the decreased serum albumin level perioperatively as an object. Additionally, the serum level of albumin was dramatically decreased after MIE in patients receiving NCT, 35 g/l as a certain cutoff for the warning signal of the AL is not as sensitive as in patients treated with surgery alone. Fortunately, decreased level of serum albumin was identified as a significant predictive aspect for the occurrence of AL in current study. Nevertheless, we still believe that it is an issue of identifying a more sensitive and accurate threshold value of serum albumin other than 35 g/l in predisposing AL occurrence in patients receiving NCT.

In our previous study, the severely decreased serum albumin level after MIE was not associated with the occurrence of AL. The plausible reasons for the conflict results may be explained as follow: [1] First, the AL rate is less than half of incidence in current study (5.8%) comparing to our previous one (11.7%) [9]. This can most likely be explained by a remarkable progression of anastomotic technique that has been achieved in our center in past years, since surgical skill was regarded as a crucial aspect affecting the occurrence of AL [2]. Secondly, in contrast with our previous study, a small population-base study with 60 patients was conducted, more than twice number of participants has been enrolled in current study. This may contribute to obtaining more accurate results [3]. Third, all patients in current study were treated with NCT, which may have a certain impact on the nutritional status but the effect on the occurrence of AL is controversial, although radiation therapy is demonstrated to be associated with the increase in AL incidence [22]. Notwithstanding, the relationship between chemotherapy and AL rate is not established due to less consistent and unequivocal evidence had been emerged. Therefore, this is one of the reasons we chose to investigate this phenomenon using patients receiving NCT with a larger study population.

Although no significant difference was observed in postoperative complications between groups, three patients suffered perioperative death (3/68, 4.4%) in MA group, whereas no mortality was observed in LA group. All of the death was suffered from AL. One was caused by the severe infection resulting from AL. The other one died of hemorrhagic shock caused by a massive hemorrhage from the benches of aortic arch (may be left subclavian artery, because no autopsy, the position of the hemorrhage cannot be identified), which was the most severe complication of AL. The last one died of ARDS attributed to the result of AL. The main cause of death in patients with AL was the severe infection in this study, which was consistent with previous research [5]. Moreover, the common characteristic of the death was hypoalbuminemia, whose serum albumin was decreased at an extremely low level postoperatively. However, whether prophylactic intravenous infusion of albumin can reduce the incidence of AL is still lack of evidence. Before surgery, nutrition administration for patients receiving NCT may contribute to improving the tissue healing [23]. Once AL occurred, closure and coverage of leakage and fluid drainage are the principles of the management strategy.

Limited data had been provided to reveal the survival difference in patients with different serum albumin levels. In current study, survival benefit was not exhibited in patients with less perioperative decreased serum albumin level after a median period of 22 months follow-up. One of the reasons may be that serum albumin level in most patients would recover to normal status after discharge, regardless of decreased level. The hypoalbuminemia may influence the physiology of the patients only in an acute phase, after recovering, deleterious effects that impair clinical outcomes may disappear. Nevertheless, the follow-up is still ongoing, and whether survival difference existing remains need to be investigated.

Although knowledge of risk factors is increasing, the prioritization of them for the development of AL is not well defined. As a quick and simple approach, detection of serum albumin, may contribute to predicting the occurrence of AL. In this study, we demonstrated that perioperative decreased serum albumin level was associated with the occurrence of AL but did not impair the survival benefit in ESCC patients treated with NCT followed by MIE. However, limitations still exist in current study. First, the number of participants was not large enough. Second, this study was conducted retrospectively in a single center. Finally, to obtain the homogeneity of the data, only ESCC patients treated with NCT followed by MIE were enrolled. Whether current conclusion can be generalized remains unknown. Therefore, further study in wider populations is warranted to validate the efficacy of perioperative decreased serum albumin levels in esophageal cancer patients. Nevertheless, gaining insight into the efficacy of prevention strategies for AL is crucial. Meanwhile, from current study, the extremely decreased serum albumin level after MIE was indicative of a warning signal enabling surgeons to prevent the AL occurrence, which is a severe and potentially life-threatening complication after MIE.