N-acetylcysteine for the prevention of stricture after circumferential endoscopic submucosal dissection of the esophagus: a randomized trial in a porcine model

Endoscopic resection of Barrett’s esophagus with a high-grade dysplasia or early esophageal cancer has become an acceptable therapeutic option [1–3], alongside surgical esophagectomy. Endoscopic mucosal resection is easily feasible but leads to a piecemeal, non-carcinologic resection of neoplastic tissue, whereas endoscopic submucosal dissection (ESD) is more carcinologically efficient, although more technically challenging. Nonetheless, both techniques result in cicatricial fibroinflammatory esophageal strictures when 75% or more of the mucosal circumference is removed [4, 5]. These strictures may require numerous endoscopic dilations, jeopardizing the patients’ quality of life and risking esophageal perforation. This is believed to result from local inflammatory reactions, as well as subsequent cicatricial fibrosis [4, 6–8]. Furthermore, pathological wound healing and fibrogenesis in the esophagus has been linked to the production of reactive oxygen species [9].

We developed an endoscopic submucosal dissection technique in a swine model, allowing for quick and safe en bloc circumferential resection of the esophageal mucosa [10]. A similar technique has been described since [11]. This method consists of a classical endoscopic submucosal dissection and blunt mechanical dissection using an endoscopic mucosectomy cap. It allows for en bloc resection of mucosal and submucosal cylinders up to 15 centimeters, without taking longer or causing more immediate complications than endoscopic mucosal resection alone. However, tight oesophageal strictures always develop in this model between the eighth and the fifteenth postoperative day.

N-acetylcysteine (NAC) is an antioxidant molecule with antifibrotic properties [12, 13], mainly related to the inhibition of transforming growth factor-beta (TGF-β) signaling [14]. It has an anti-inflammatory effect caused by downregulating TNF-α, IL-6, and IL-8 synthesis [15]. Furthermore, it is a cheap, readily available, and a safe molecule. NAC has shown efficiency in pulmonary fibrosis [16] and is of possible interest in cirrhotic [17], nonalcoholic steatohepatitis [18] liver fibrosis, and improved wound healing of tympanic membranes [19]. In the esophagus, anastomotic wound healing improved with NAC administration [20].

Therefore, we conducted a controlled trial to assess the interest of NAC in the prevention of esophageal strictures, following circumferential ESD in our swine model.

The median esophageal diameter on the 21^st postoperative day was 4 mm (range 2 to 5) in group 1 (treated with NAC) and 3 mm (range 1 to 7) in group 2 (control) (P = 0.95) (Figure 1 and 2).

In group 1, none of the six animals exhibited abnormal behavior, but five out of six (83%) were unable to ingest a regular diet after fourteen days because of regurgitations. In group 2, one pig out of five (20%) was found to be asthenic, and four out of five (80%) were unable to eat because of their esophageal stricture. Overall number of dilation was 13 (2.2 per pig) in group 1 and 12 (2 per pig) in group 2 on day 21 (P = NS).

Histology revealed two different patterns that were found indistinctively in both groups. In the first one (Figure 3), the superficial layer remained ulcerated; we observed a dense granulation tissue with a major inflammatory cell infiltrate and high cell density with predominance of polynuclear leukocytes in the remaining submucosa; we also observed myofibroblasts in the submucosal and muscular layers, surrounded by fibrosis (Figure 3, A2 and B2). Conversely, we observed a neoepithelium in the second pattern (Figure 4), characterized by a single layer of immature epithelial cells. Fibrosis (Figure 4, A2 and B2) was found underneath the neoepithelium, around myofibroblasts and fibroblasts, filling the submucosal layer and dissecting muscular fibers; associated with moderate, lymphoplasmocytic inflammatory cell infiltrate, and granulation tissue (Figure 4, A and B). These two patterns can be hypothesized as corresponding to the early and late phases of the same inflammatory and fibrotic cicatricial process. Median fibrosis thickness was 1740 μm (range 1030 to 2570) in group 1 vs 1915 μm (range 1710 to 2860) in group 2, P = NS; median thickness of the granulation tissue was 1000 μm (range 925 to 1100) in group 1 vs 843 μm (range 868 to 2440) in group 2, P = NS. The granulation tissue presented with features of acute inflammation in 30% of cases in group 1 vs 25% in group 2. Immunohistochemistry staining with 8-hydroxy-2’ deoxyguanosine (8-OH DG) antibody showed a more intense signal in the esophageal wall of control animals as compared to NAC-treated animals (Figure 5). This finding was consistently observed on all slides. However, signal measurements did not show a statistically significant difference between the groups: median: 148 (range 122 to 169) in group 1 vs 136 (range 90 to 168) in group 2, P = 0.54.

This study is the first one to evaluate, on a swine model, the systemic administration of an antioxidant molecule in the prevention of esophageal strictures.

Nonetheless, this work did not demonstrate any clinical, endoscopical, or histological benefit of NAC to prevent oesophageal stenosis after circumferential ESD. A first hypothesis is an insufficient oral intake of NAC in regurgitating animals, despite repeated esophageal dilations. However, if solid food intake is impaired by esophageal stricture, liquid intake is almost always possible. Moreover, food intake remained normal during the first 10 days, the period during which NAC was expected to exert its maximal beneficial effect, if any. An intravenous route for NAC would have required daily invasive treatment administration, and sedation of the animals, which did not seem acceptable; furthermore, oral and intravenous administrations have been reported to be equivalent [21]. Doses of NAC vary from 20 mg/kg to 400 mg/kg of body weight per day. Given that the bioavailability of orally administrated NAC is low (<10%), we decided to use high doses, closer to those used in acetaminophen poisoning than in chronic fibrotic diseases [22]. The pigs were put on a 60 mg/kg/day dose of NAC as part of a preliminary experiment and results showed improvement of the esophageal diameter at day 15 with no side effects. This propelled us to increase the NAC dose to 100 mg/kg/day. The use of higher doses, up to 400 mg/kg/day, has been reported, but is commonly associated with frequent side effects [23]. Such anaphylactoïd reactions, including vomiting, could have biased the evaluation of food intake, esophageal mucosa, and symptoms related to esophageal stricture. We did not record such anaphylactoïd reactions in the animals we treated. However, insufficient local concentration could also explain these results and could warrant another attempt of NAC administration, directly into esophageal wounds.

The second issue is the timing of the clinical and histological assessment. We chose the 21^st postoperative day as the day of endpoint assessment because it was consistent with the timing of stricture development in this model, generally observable between days 10 and 15 after ESD. Animals were kept in good condition after day 15, using iterative endoscopic dilations. Of note is the fact that these dilations might have played a role in the delayed reepithelialisation observed in some animals. It can be hypothesized that an earlier assessment of clinical and histological endpoints might have shown a more significantly different data, but the absence of any observable benefit at 21 days suggests such findings would not have been highly relevant to further clinical use of NAC.

A third hypothesis for NAC inefficiency is the severe and systematic nature of oesophageal strictures in pigs (possibly due to continuous acid aggression in conditions of stress). This makes our swine model particularly heuristic for the study of stricture development and prevention: if a treatment provides positive clinical results in this model, it should be all the more efficient in humans, but one would expect that NAC should at least exhibit some histological effect.

NAC has been found to have a direct antioxidant effect as well as to stimulate glutathione synthesis, a major endogenous intracellular antioxidant. It also plays a role in modulating the proinflammatory cytokine response [24], notably by inhibiting TGF-β1 [12] and abrogating the fibrogenic properties of pathological fibroblasts [14]. Post-endoscopic esophageal strictures are characterized by an intense local inflammatory reaction, which prevents reepithelialisation taking place and quickly makes way for cicatricial fibrosis. An increased level of reactive oxygen species (ROS) has been reported in esophageal ulcers, with the benefit of an antioxidant molecule (vitamin E) administration on esophageal wound healing [9]; along with these results, many stricture-preventing drugs, either antioxidant [25], antifibrotic [26–30], or anti-inflammatory [4, 7, 31], have been tested on esophageal wounds. The only treatment reported to be beneficial against esophageal strictures in human studies is local corticosteroid injection [4], although these results are contradictory, with previous animal studies reporting severe complications related to triamcinolone injection [7]. These data have led us to consider NAC for post-endoscopic esophageal stricture prevention. Despite an antioxidant effect suggested by the results of immunohistochemistry (although the difference between both signals on quantitative measurements did not reach statistical significance), NAC did not, in our work, prevent esophageal stricture.

Considering our results, it seems that several mechanisms could be at stake in cicatricial esophageal fibrosis. In particular, mucosal defect and resulting acid, and bacterial aggressions of the submucosa could play a role; therefore, stricture prevention strategies focusing on covering and protecting the injured esophagus should be developed, even if recent studies, including ones done on humans, either report insufficient efficacy [32], or show an improvement of esophageal wound healing only in partial mucosal defects [7, 8, 33–35], where strictures are uncommon. However, the use of growth factors, extracellular membranes, or bio-scaffolds or cellular therapy, makes their translation into daily practice unlikely for the present moment. This data confirm that preventive treatment should prevail against curative ones regarding post-endoscopic esophageal strictures.

Our study failed to prove that an antioxidant such as NAC can improve the outcome of tissue repair in a model of severe post-ESD esophageal stricture, when used as a single agent. Its antioxidant activity did not prevent fibroinflammatory cicatricial esophageal stricture formation at the 21st postoperative day. Extensive esophageal endoscopic submucosal dissection might become an alternative choice to esophagectomy for Barrett’s esophagus with high-grade dysplasia or early adenocarcinoma; hence, further analysis of esophageal wound healing will be needed, in order to develop new stricture-preventing treatments.