Late-onset thoracogastric airway fistula after esophagectomy for esophageal cancer: a narrative review

Abstract

Background and Objective

Thoracogastric airway fistula (TGAF) is a life-threatening complication following esophagectomy with gastric conduit interposition. While early-onset TGAF has been well characterized, late-onset TGAF, which occurs months to years after surgery, remains underrecognized and lacks evidence-based diagnostic and management frameworks. This narrative review aims to synthesize current literature to address several high-priority clinical questions: which features should raise suspicion of lateonset TGAF; what is the optimal diagnostic pathway integrating imaging and endoscopy; and how to select between endoscopic and surgical intervention.

Methods

We conducted a systematic search of PubMed from database inception to May 2026 using combinations of keywords related to TGAF and relevant etiological and therapeutic terms. Reference lists of included articles and relevant reviews were manually screened. Only peer-reviewed English-language publications were considered. Given the rarity and heterogeneity of late-onset TGAF, high-quality prospective studies are lacking, and a systematic review with meta-analysis is not feasible. We therefore adopted a narrative synthesis approach.

Key Content and Findings

The evidence base for late-onset TGAF is limited to case reports, small series, and a few cohort studies, with no randomized trials identified. Diagnostic delay is common, driven by nonspecific symptoms and overlap with tumor progression. The pathogenesis is multifactorial, with peptic ulceration as the bestsupported mechanism. Management options include conservative support, endoscopic stenting or occluder devices, and surgical repair, though treatment decisions are guided largely by patient physiology, fistula characteristics, and institutional expertise rather than by comparative data. Long-term outcomes remain poorly characterized.

Conclusions

This review presents an initial framework for managing late-onset TGAF. A high index of suspicion should be maintained in any post-esophagectomy patient presenting with recurrent pneumonia, chronic cough, or unexplained weight loss. Diagnostic workup should proceed with contrast esophagram and chest computed tomography (CT), followed by confirmatory bronchoscopy and gastroscopy. Endoscopic stenting is appropriate as firstline therapy in most patients, whereas surgical repair should be reserved for those with adequate physiological reserve and favorable local tissue conditions. The limited evidence base underscore the urgent need for prospective multicenter registries to establish standardized diagnostic algorithms and treatment guidelines for this rare but devastating complication.

Introduction

Esophageal cancer is among the most prevalent malignancies of the digestive tract globally (1,2), and esophagectomy with reconstruction using a gastric conduit is the preferred surgical approach (3,4). Despite advances in surgical technique and perioperative care, postoperative complications remain an unavoidable clinical challenge. Among them, thoracogastric airway fistula (TGAF) is a rare but life-threatening complication characterized by a pathological communication between the thoracostomach and the tracheobronchial tree (5), with a reported incidence ranging from 0.3% to 1.9% (5-10). The continuous reflux of acidic gastric contents through the fistula into the airway induces aspiration pneumonia, respiratory failure, and sepsis, resulting in a disproportionately high mortality rate among confirmed TGAF cases (9,11).

Based on temporal onset, TGAF is conventionally classified into early-onset and late-onset subtypes in the literature; however, this dichotomy remains unincorporated into standardized clinical frameworks. Early-onset TGAF, typically manifesting within the first few weeks postoperatively, exhibits a relatively well-defined pathophysiology and clinical presentation, enabling well-established, evidence-based diagnostic criteria and therapeutic protocols (8,12,13). In contrast, the evidence base for late-onset TGAF, defined as fistula formation occurring months to years after initial esophagectomy (8,10), remains limited and fragmented. Its low incidence, marked clinical heterogeneity, and prolonged latency period impede prospective cohort studies and multicenter validation, leaving several high-priority clinical questions unresolved. First, because late-onset TGAF often presents with non-specific symptoms that mimic tumor recurrence, post-radiation changes, or community-acquired pneumonia, there are no validated alert signs or screening strategies to support early recognition during routine follow-up. Second, the optimal diagnostic pathway, particularly the sequence and integrated application of imaging modalities and endoscopic evaluation, has not been prospectively defined. Third, indications for endoscopic intervention versus surgical repair remain inadequately specified, particularly in patients with prior radiation, malnutrition, or tumor recurrence. Fourth, prognostic determinants have not been systematically evaluated, precluding risk-stratified decision-making and personalized intervention. Although recent case reports and small case series have begun to delineate diagnostic patterns and therapeutic outcomes, a comprehensive summary of pathogenic mechanisms and evidence-based intervention strategies remains lacking (8,14,15).

Given the rarity and clinical heterogeneity of late-onset TGAF, high-quality prospective studies and randomized controlled trials remain unavailable, rendering systematic review or meta-analysis unfeasible. To address this knowledge gap, we conducted a narrative review synthesizing current evidence on late-onset TGAF, with emphasis on its temporal definition, multifactorial pathogenesis, clinical presentation and associated diagnostic challenges, and therapeutic strategies. By consolidating fragmented reports into a unified framework, this review aims to improve diagnostic vigilance and inform multidisciplinary management decisions. We present this article in accordance with the Narrative Review reporting checklist (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2026-1489/rc).

Methods

Relevant literature was identified in line with the scope of this narrative review on late-onset TGAF. We conducted a systematic search of PubMed from database inception to May 2026 using combinations of the following key terms: “esophagectomy”, “gastric conduit”, “thoracogastric airway fistula”, “gastrotracheal fistula”, “gastrobronchial fistula”, “late-onset”, “peptic ulcer”, “radiation injury”, “tumor recurrence”, “chronic ischemia”, “stent”, “occluder”, and “surgical repair”. The search was limited to English-language publications. In addition, we manually screened the reference lists of all included articles and relevant reviews to identify additional studies not captured by the primary search. Records indexed in PubMed or published in recognized peer-reviewed journals were included to ensure methodological rigor and source credibility. A single exception was made for the conference abstract by Magana et al. (14), which reports the longest documented latency interval (13 years) for late-onset TGAF. This record was cited exclusively to illustrate the potential for extreme clinical latency and was excluded from analyses pertaining to pathogenesis, clinical characteristics, or treatment outcomes. Final inclusion was determined based on relevance to the prespecified clinical themes of this review, with emphasis on clinical applicability, and availability of recent, high-quality evidence. The search strategy is summarized in Table 1.

Table 1. The search strategy summary.

Items	Specification
Date of search	Initial search: 10 March 2026; final update: 21 May 2026
Databases and other sources searched	PubMed, supplemented by screening of reference lists of relevant articles
Search terms used	A systematic literature search was conducted using combinations of the following key terms: “esophagectomy”, “gastric conduit”, “thoracogastric airway fistula”, “gastrotracheal fistula”, “gastrobronchial fistula”, “late-onset”, “peptic ulcer”, “radiation injury”, “tumor recurrence”, “chronic ischemia”, “stent”, “occluder”, and “surgical repair”
Timeframe	From database inception to 21 May 2026
Inclusion criteria	English-language articles relevant to TGAF; no restrictions on study design; peer-reviewed records indexed in PubMed or published in recognized journals
Selection process	Literature was selected based on relevance to the scope of this narrative review, with emphasis on clinical applicability, recent evidence where available, and landmark studies when appropriate

TGAF, thoracogastric airway fistula.

Evidence on late-onset TGAF is derived from heterogeneous sources, including single-case reports, small case series, and a limited number of retrospective cohort studies, with considerable variation in patient characteristics, clinical management approaches, and reported outcomes. To synthesize these disparate data, we prioritized evidence from studies with larger sample sizes or more rigorous methodological designs where available, while recognizing that case-level reports often constitute the sole source of insight into rare pathophysiological mechanisms.

Temporal classification and pathogenesis

Distinguishing between early- and late-onset TGAF is critical, as the two subtypes differ not only in temporal onset but also in underlying pathogenic mechanisms and clinical phenotypes. Early-onset TGAF is conventionally defined as fistula formation within the first few weeks postoperatively. Prospective observational multicenter cohort indicate that the pathogenesis of this condition is primarily attributable to perioperative insults, including thoracogastric ischemia or necrosis, technical deficiencies at the anastomotic site, and iatrogenic tracheobronchial injury secondary to endotracheal intubation (16-18). Yasuda et al.’s small case series (n=10) proposed a refined subclassification: anastomotic leakage-associated TGAF (onset 8-35 days; n=5) and gastric necrosis-associated TGAF (onset 20–36 days; n=4) (8). This evidence remains foundational for delineating the early-onset window and underscores the primacy of surgical technique and regional perfusion in driving the pathogenesis of early-onset TGAF.

Late-onset TGAF is defined by a fistula that presents at least three months after esophagectomy (10). To date, a single case report by Magana et al. (14) represents the longest documented latency interval (13 years) for late-onset TGAF, tentatively supporting its characterization as a delayed complication that may remain clinically quiescent for more than a decade. However, it is important to recognize that this single-case observation, while highly instructive, does not by itself establish the frequency or generalizability of such extreme latency. The pathogenesis of late-onset TGAF does not represent a delayed recurrence of early-onset mechanisms. Instead, it results from the cumulative effect of interdependent chronic processes, including peptic ulcer-induced mucosal erosion, radiation-induced fibroatrophic tissue remodeling, recurrent neoplastic infiltration, chronic thoracogastric ischemia, and persistent mechanical stress at the conduit-airway interface, a profile fundamentally distinct from that of early-onset TGAF.

Peptic ulcer

Peptic ulceration of the thoracostomach is considered a clinically relevant cause of late-onset TGAF, supported by multiple lines of indirect and direct evidence. Endoscopic surveillance cohort studies have reported a thoracogastric ulcer prevalence of 6.1% to 19.4% among long-term survivors following esophagectomy (19,20), indicating that ulceration is not uncommon in this population. Anatomically, as a transmural ulcer extends through the gastric conduit wall, it may erode into the adjacent tracheal or bronchial wall, thereby establishing a direct fistulous communication between the gastric lumen and the airway tree, a progression described in several case reports (8,21-24). While these observations provide proof of principle that chronic ulcerative disease can precipitate fistula formation, they do not quantify the incidence of this pathogenic pathway due to the paucity of reported cases.

Additional risk factors, including Helicobacter pylori infection and chronic use of nonsteroidal anti-inflammatory drug (NSAID), have been independently associated with impaired gastric mucosal defense and peptic ulcer development. In the general population, H. pylori infection is implicated in approximately 42% of peptic ulcer cases, while NSAID use accounts for roughly 36% (25). The pathogenic mechanisms of NSAID-induced injury are relatively well understood, involving both direct mucosal toxicity and suppression of prostaglandin synthesis, which plays a key role in maintaining mucosal integrity, microvascular perfusion, mucus and bicarbonate secretion, and epithelial restitution (26). The coexistence of H. pylori infection and NSAID use acts synergistically to promote ulcer formation, substantially amplifying ulcer risk, particularly among patients with preexisting mucosal vulnerability or other comorbidities (27). However, it is important to recognize that most of this evidence comes from studies of the native stomach, and its direct applicability to the denervated, ischemic thoracic gastric conduit remains uncertain. Thus, while H. pylori infection and NSAID use represent plausible contributing factors to conduit ulceration, their precise role in the pathogenesis of late-onset TGAF requires further investigation.

Radiation-induced injury

Radiation therapy is recognized as a potential contributor to late-onset TGAF, primarily through its well-documented effects on microvascular integrity and tissue repair capacity. In the broader radiation biology literature, ionizing radiation has been shown to induce a cascade of microvascular injury, progressing from endothelial damage and capillary rarefaction to chronic tissue hypoperfusion, ischemic fibrosis, and ultimately structural impairment (28-30). This cascade exhibits substantial latency, with clinical manifestations typically emerging months to years after radiotherapy completion, well beyond the timeframe of acute or subacute radiation injury. Several case reports have described radiation-associated bronchoesophageal fistula formation occurring seven months after radiotherapy (31), and gastric conduit necrosis with anastomotic leakage in a patient undergoing esophagectomy as late as seven years following prior gastric irradiation (32). These observations suggest that radiation-damaged tissues exhibit profoundly impaired regenerative capacity; consequently, even minor secondary insults, such as microbial colonization, localized inflammation, or mechanical stress, may trigger transmural necrosis and perforation at the conduit-airway interface (32).

However, several limitations must be acknowledged when interpreting these data. First, the evidence for radiation-induced TGAF rests predominantly on case-level observations rather than systematic cohort studies. Second, in many reported cases, radiation injury coexists with other potential contributory factors, such as tumor recurrence or surgical trauma, making it difficult to isolate the specific contribution of radiation. Despite these uncertainties, the available evidence supports maintaining a high diagnostic consideration for radiation-induced late-onset TGAF in patients with a history of postoperative thoracic radiotherapy, not only during early surveillance but also throughout long-term follow-up.

Tumor recurrence

Locoregional tumor recurrence represents a clinically important oncologic mechanism underlying late-onset TGAF, particularly when it involves the esophagogastric anastomosis or the intrathoracic gastric conduit wall. The pathogenic rationale is biologically straightforward: recurrent neoplastic tissue may directly infiltrate and erode through the full thickness of both the conduit wall and the adjacent tracheobronchial wall, thereby establishing a pathological fistulous tract. This mechanism has been observed in surgical series of post-esophagectomy fistulas (33) and clinical observations showing that tumor-associated TGAF typically co-occurs with signs of disease progression, including progressive dysphagia, intractable retrosternal or interscapular pain, and serial elevation of tumor markers (34). Cross-sectional imaging, especially contrast-enhanced computed tomography (CT) and positron emission tomography (PET)/CT, frequently demonstrates a metabolically active, irregular soft-tissue mass or asymmetric, nodular wall thickening with heterogeneous enhancement (35). The emergence of a fistula in this context often indicates advanced disease and is strongly associated with limited therapeutic options and a poor prognosis (36), which is consistent with the general oncologic principle of local progression, but systematic data comparing outcomes by etiology are lacking. Nonetheless, in patients with known or suspected locoregional recurrence, the possibility of TGAF should be actively considered, particularly when new respiratory or swallowing symptoms develop.

Chronic ischemia and mechanical stimulation

In addition, chronic ischemia and mechanical stimulation act synergistically to promote late-onset fistula formation. Following esophagectomy with gastric conduit reconstruction, the thoracic gastric conduit relies predominantly on the right gastroepiploic artery (RGEA) for perfusion, a vessel with an elongated course and inherently low distal perfusion pressure (37,38). Compromise of the RGEA can result in ischemia or necrosis (38,39), or it can leave the conduit wall in a state of chronic, borderline ischemia (40). Concurrently, prolonged apposition between the thoracic gastric conduit and the membranous portion of the trachea, coupled with repetitive friction and compression from respiratory excursions and cardiac pulsation, imposes cumulative mechanical stress on the gastric wall. The ischemic milieu reduces the gastric wall’s tolerance to injury, while persistent mechanical stimulation continuously acts on vulnerable areas. Together, these factors promote degenerative changes in the gastric wall and may ultimately lead to perforation triggered by an inciting event.

In summary, the pathogenesis of late-onset TGAF is multifactorial, involving peptic ulceration, radiation-induced tissue injury, tumor recurrence, and chronic ischemia and mechanical stimulation. These mechanisms may act independently or synergistically. The evidence supporting each of these mechanisms varies in strength: peptic ulceration is supported by both endoscopic surveillance data and direct case-level observations of ulcer perforation; radiation injury is well established in the broader literature on radiation-induced tissue damage, though direct evidence in the context of TGAF remains limited; tumor recurrence is a well-recognized oncologic principle; while the roles of chronic ischemia and mechanical stress, though biologically plausible, rest largely on indirect evidence and anatomical reasoning. Therefore, a systematic evaluation that integrates temporal factors, treatment history, and oncologic status is essential for establishing the diagnosis.

Clinical manifestations

Typical symptoms of TGAF include postprandial choking, recurrent pneumonia, intermittent expectoration of ingested food, and systemic wasting (5,41-43). However, the onset pattern and symptom severity differ significantly between early-onset and late-onset cases.

Early-onset TGAF typically manifests within weeks after surgery, often while the patient remains hospitalized or is under close surveillance shortly after discharge. Its clinical presentation is characteristically acute and rapidly progressive. The most distinctive feature is the abrupt onset of severe coughing immediately after oral intake. Because this symptom is temporally linked to eating, it reliably raises high clinical suspicion in both patients and clinicians. In addition, patients with early-onset fistulas frequently exhibit fever, expectoration of food debris, recurrent aspiration pneumonia, and acutely progressive systemic wasting. Owing to this classic symptom presentation and the heightened diagnostic vigilance during the perioperative period, early-onset TGAF is usually diagnosed promptly, with a low rate of missed diagnosis.

In contrast, although late-onset TGAF shares similar symptoms with its early-onset counterpart, their presentation is often atypical and insidious, which commonly leads to diagnostic delay or misdiagnosis. The principal clinical features and their associated diagnostic pitfalls are outlined below.

Postprandial cough

Patients with late-onset TGAF may also experience postprandial cough; however, the clinical pattern differs meaningfully from that observed in early-onset disease. In early-onset TGAF, cough occurs immediately after oral intake, is severe, and is readily recognized as abnormal. In late-onset TGAF, cough may develop insidiously, be milder in intensity, or be selectively triggered by specific consistencies, such as liquids or semi-liquids. These subtle presentations are frequently misinterpreted by both patients and clinicians as “chronic cough” or “throat discomfort”, thereby lowering diagnostic suspicion for TGAF (44,45). Consequently, patients may undergo repeated evaluations for common causes of chronic cough without identification of the underlying fistula.

Recurrent pneumonia

Patients with late-onset TGAF frequently develop aspiration pneumonia secondary to recurrent microaspiration of gastrointestinal contents into the airway. Unlike those with early-onset TGAF, who typically develop pneumonia during hospitalization where it is readily identified, patients with late-onset TGAF often present repeatedly to outpatient clinics with a clinical course mimicking community-acquired pneumonia, characterized by cough, sputum production, and fever (11,46). Although symptoms may transiently improve with antibiotic therapy, they inevitably recur due to persistent fistulous communication. This clinical pattern frequently leads clinicians to treat each episode as isolated pneumonia without investigating the underlying structural pathology. Consequently, patients may undergo multiple courses of antibiotics and repeated chest imaging, delaying the correct diagnosis until more specific signs, such as expectoration of food particles, eventually emerge.

Expectoration of food debris

Expectoration of food debris is a highly specific manifestation of TGAF. However, in patients with late-onset TGAF, this symptom often occurs intermittently. Patients may notice food particles in their sputum only after ingestion of specific food consistencies, such as liquids or soft solids. Some patients even attribute the phenomenon to “pharyngeal foreign body sensation” or “food residue in sputum”, delaying medical evaluation. Moreover, because the fistula opening may be small, formed food particles are less likely to pass through, whereas liquids or semi-liquids flow more easily into the airway. This selective passage explains why patients may present with intermittent or consistency-dependent food expectoration rather than a constant symptom.

Systemic wasting

Patients with late-onset TGAF frequently present with chronic, progressive systemic wasting, evidenced by weight loss, malnutrition, and hypoalbuminemia. Clinicians may readily misattribute this wasting to primary tumor progression or recurrent metastasis (47), prioritizing oncologic evaluation while overlooking TGAF as a potentially reversible and non-neoplastic complication, thus leading to diagnostic delays. This diagnostic bias is further reinforced by the absence of typical TGAF symptoms, such as postprandial choking or food expectoration. Consequently, patients may undergo repeated imaging studies and even unnecessary changes in oncologic treatment regimens before the correct diagnosis is established.

Management approaches

Current treatment strategies for late-onset TGAF fall into three categories: conservative management, endoscopic/interventional therapy, and surgical treatment (48). Given that late-onset TGAF frequently arises in the context of adverse local tissue conditions, such as dense adhesions, progressive fibrosis, loss of anatomical landmarks, and compromised vascularity, and is often compounded by systemic factors including malnutrition, prior radiation injury, or tumor recurrence, therapeutic decision-making must be carefully individualized. Clinical selection among these options should be guided by a systematic, multidimensional assessment encompassing fistula characteristics (e.g., diameter and tract length), the patient’s functional status and comorbid burden, the underlying etiology, and the presence or absence of active infection.

Conservative management

Conservative management is indicated for patients with significant comorbidities or frailty who are unsuitable for endoscopic or surgical intervention, as well as for the rare subset of patients with minute fistulas demonstrating radiographic or endoscopic evidence of spontaneous closure potential. Its primary objectives are threefold: infection control via targeted antimicrobial therapy and, when indicated, closed thoracic drainage; interruption of gastrointestinalairway communication through fasting, gastric decompression, and acid suppression with proton pump inhibitors; and nutritional rehabilitation via enteral (e.g., nasojejunal tube or jejunostomy) or parenteral nutrition when enteral access is not feasible (11,49).

Notably, conservative management is associated with a substantial rate of intervention failure. In the largest available study (Zhang et al., 212 patients), 92 received conservative management, including the two-tube method for nutritional rehabilitation, as sole treatment because they were unsuitable for surgery or stenting. Mortality in this subgroup was 47.8% (44/92), attributed to pulmonary infection, bleeding, or tumor progression (11). These figures indicate that conservative management, when used definitively, carries high failure rates, particularly in patients with advanced disease or poor physiological reserve. The authors themselves described this approach as “a bridge for successive treatments” rather than a cure, reinforcing its temporizing role. In the absence of standardized selection or escalation criteria, the decision to pursue conservative management should be individualized, with explicit acknowledgment of its limited efficacy and the risk of disease progression during conservative trials.

Endoscopic/interventional therapy

Advances in endoscopic and image-guided interventional techniques have established endoscopic and interventional therapy as the preferred initial treatment for late-onset TGAF, particularly in patients at high surgical risk or those who decline surgical intervention. Compared with surgical repair, this approach offers several well-documented advantages, including minimal procedural trauma, shorter recovery time, and the feasibility of repeat application. Available interventional modalities include airway stent placement, esophageal or gastric stent placement, occluder device closure, and endoscopic suturing or clipping.

Airway stent placement

Covered airway stents represent the most widely adopted endoscopic intervention for TGAF (50). Under bronchoscopic guidance, a covered metallic or silicone stent is deployed in the trachea or main bronchus to occlude the fistulous tract and prevent aspiration of gastrointestinal contents. For fistulas involving the carina or main bronchi, a Y-shaped covered stent provides conformal sealing. In complex cases where a single Y-stent fails due to large or irregular fistulous tracts, a combined Y-en-Y technique may be employed: a small Y-stent is first deployed into the distal part of the involved airway, followed by a large Y-stent placed at the trachea and carina, with the large stent partly overlapping the main body of the small stent. This sequential placement achieves thorough, multilayer closure of the airway defect (51).

Airway stents include straightforward procedural execution and reliable immediate occlusion. The reported immediate technical success rate for airway stenting was 93.65% in a case series of 63 patients reported by Han et al. (50). However, long-term outcomes are less favorable. A retrospective analysis of 58 TGAF patients treated with airway stents, reported by Wang et al., documented a healing rate of only 34.5%, although complete sealing was achieved in 77.6% (52). This discrepancy highlights an important distinction: radiographic sealing does not necessarily translate into durable fistula healing. Long-term complications remain significant, including stent migration, granulation tissue-induced airway restenosis, sputum retention, and recurrent respiratory infections. In Wang’s cohort, all patients developed retention of secretions and varying degrees of granulation tissue formation, with stent migration occurred in three cases (5.2%). During followup (range, 2–61 months), 30 patients died of organ failure (mean 8.4±4.3 months), 16 of respiratory failure (mean 11.0±9.3 months), and two of massive hemoptysis (52), underscoring the poor overall prognosis despite procedural success.

Thus, while airway stenting offers an effective minimally invasive option for fistula occlusion, its longterm efficacy is limited by incomplete healing and device-related complications. Patient selection and postprocedural surveillance remain critical, particularly in late-onset TGAF.

Esophageal and gastric stent placement

In patients with a large gastric-side fistula or persistent leakage following airway stent placement, deployment of a covered stent within the thoracic gastric conduit may be considered to achieve occlusion from the gastrointestinal side. However, the anchoring efficacy of gastric stents is inherently lower than that of airway stents due to the larger luminal diameter and peristaltic activity of the gastric conduit (51).

In select complex cases, such as those with wide fistulous tracts, irregular margins, or failed single-stent occlusion, a dual-stent technique may be employed. This approach involves concurrent placement of covered stents in both the airway and the gastrointestinal tract to achieve circumferential sealing. While theoretically appealing, the evidence supporting this strategy is limited to small case series, and the technique carries higher procedural complexity and increased complication risks, including stent-stent friction, erosion, and airway compromise (53,54). Published experience does not permit reliable conclusions on superiority over single-stent strategies, and optimal patient selection criteria remain undefined. Given these uncertainties, dual-stent placement should be used cautiously, with careful individualized assessment of fistula anatomy and patient-specific risk factors.

Occluder device closure

Cardiac septal defect occluders, including atrial septal defect (ASD) and ventricular septal defect (VSD) occluders, have been increasingly repurposed for the endoscopic management of TGAF. The device’s bilateral self-expanding disk configuration is deployed across the fistula, achieving simultaneous closure from both the airway and thoracostomach sides (55). Unlike covered airway stents, occluders preserve airway patency by avoiding intraluminal obstruction, thus maintaining mucociliary clearance and sputum drainage, and circumvents well-documented stent-related complications. Emerging evidence from case reports and small case series supports the technical feasibility and short-term safety of this approach, with accurate measurement of fistula dimensions is pivotal to its success (56,57).

However, several challenges limit broader application. The anatomical and histological differences between cardiac septal defects and TGAF mean that occludertissue fitting may be suboptimal (58). Successful application hinges on operator expertise in advanced interventional endoscopy and requires a fistula anatomy characterized by well-defined margins and relatively regular geometry, excluding many patients with complex or irregular fistulas (59). Moreover, comparative studies evaluating longterm outcomes against conventional stenting remain lacking, and optimal patient selection criteria have yet to be defined.

Surgical treatment

Surgical intervention represents the ultimate radical treatment for late-onset TGAF; however, it is also associated with the highest risk and greatest trauma. Primary surgical indications include failure of endoscopic or interventional management, large fistulas refractory to occlusion by endoscopic or device-based techniques, and locoregional tumor recurrence requiring concurrent resection. Preoperatively, rigorous infection control and comprehensive nutritional optimization are essential. Bronchoscopy and gastroscopy should be performed to precisely delineate fistula location, dimensions, and perifistula tissue viability. Contrast-enhanced chest CT is essential to evaluate fistula proximity to mediastinal vasculature and adjacent critical structures, guiding surgical strategy and approach selection.

Direct repair or with tissue flap interposition

In patients with favorable local tissue viability and no evidence of tumor recurrence, direct fistula repair may be considered. However, owing to compromised regional perfusion and impaired wound-healing capacity in the irradiated or chronically inflamed tissue bed, primary suture closure alone is associated with high failure rates (60). Consequently, interposition of a well-vascularized pedicled tissue flap is often used to restore local blood supply, obliterate the conduit-airway communication, and augment structural integrity. Commonly used flaps include the pectoralis major flap (61), latissimus dorsi flap (62), pericardial flap (60), as well as perforator flaps such as the internal mammary artery flap and dorsal intercostal artery flap (10). Surgical access is determined by fistula location and anatomy, with options including posterolateral thoracotomy, median sternotomy, or cervical incision (10,60,62,63).

The reported outcomes of tissue flap repair are derived from small, heterogeneous case series, and the available data do not permit reliable comparisons between different flap types or surgical approaches. Most publications describe technical feasibility rather than long-term functional outcomes, and the criteria for selecting one flap over another remain largely experience-based rather than evidence-driven. Despite these evidence gaps, tissue flap interposition remains an important surgical option in carefully selected patients, and its use should be guided by individual anatomical considerations and surgical expertise rather than by high-level comparative evidence, which is currently lacking.

Resection of the thoracostomach with colonic or jejunal interposition

In patients with extensive ischemic necrosis of the thoracostomach, multiple refractory ulcers, or confirmed tumor recurrence, direct repair is inadequate. Definitive management requires resection of the compromised thoracostomach and reconstruction of the alimentary tract using either colonic or jejunal interposition for esophageal replacement (64-68). Although these procedures carry significant surgical trauma, this is the definitive management when no other reconstructive options exist. Preoperative evaluation must rigorously assess both systemic physiological reserve and mesenteric vascular adequacy to ensure patient suitability for this major procedure.

The evidence base for colonic or jejunal interposition in salvage esophageal reconstruction is derived from multiple case series and systematic reviews, confirming its technical feasibility in experienced centers. Although none specifically addressed the unique challenges of late-onset TGAF, including local ischemia, fibrosis from prior radiation, or tumor recurrence, the principle of replacing the compromised gastric conduit with healthy, well-vascularized tissue offers a sound rationale for overcoming unfavorable local conditions. In the colonic interposition series, overall survival at 5 and 10 years was 33% and 25%, respectively, for patients with esophageal cancer (64). For the SPJ cohort, the 5‑year survival rate was 30% (68). Given the complexity of these procedures and the heterogeneity of the patient population, careful patient selection in high‑volume centers with multidisciplinary expertise is essential to optimize outcomes.

Tracheal or bronchial sleeve resection and reconstruction

When the tracheal or main bronchial wall adjacent to the fistula exhibits extensive necrosis or confirmed tumor invasion, primary repair is unlikely to be durable. In such cases, sleeve resection of the affected airway segment, followed by tension-free end-to-end anastomosis, is indicated to restore anatomical continuity and functional airway integrity (69).

Tracheal and bronchial sleeve resection is technically mature, though complications remain a significant concern. Overall morbidity occurs in approximately 20% of cases, half of which are anastomotic complications (70). Other reported complications include granulation tissue formation, restenosis, tracheoesophageal or tracheoinnominate fistula, wound infection, laryngeal edema, and glottic dysfunction (71,72). In patients with late-onset TGAF, additional risk factors, including active infection, chronic inflammation, malnutrition, hypoalbuminemia, and prior radiation therapy, compound the risk by impairing wound healing and tissue viability, further predisposing to dehiscence and stenosis. Therefore, sleeve resection should be reserved for highly selected patients with acceptable nutritional status and physiological reserve, and performed in centers with multidisciplinary expertise in complex airway reconstruction and perioperative nutritional support (73).

In summary, management of late-onset TGAF requires individualized, physiology-driven decision-making informed by comprehensive clinical and anatomical assessment. In most patients, particularly those with poor physiological reserve, active infection, or advanced tumor burden, endoscopic stenting serves as the appropriate firstline therapy, offering rapid symptom relief with acceptable risk. For patients with small, regularly shaped fistulas without active infection or necrosis, endoscopic or interventional therapy remains the initial modality of choice. Surgical repair should be reserved for those with adequate physiological reserve and favorable local tissue conditions, and prioritized for large or anatomically complex fistulas, especially those near mediastinal vasculature, to achieve durable anatomical restoration. In patients with concomitant locoregional tumor recurrence, oncologic control takes precedence, and surgical planning must be guided by precise staging. Following interventional failure or recurrence, the feasibility of repeat intervention versus surgical conversion should be carefully reassessed. Given the rarity and heterogeneity of TGAF, high-quality prospective evidence remains scarce, and treatment decisions should be made on a case-by-case basis within a multidisciplinary framework.

Conclusions

Compared with early-onset fistulas, late-onset TGAF has received markedly less attention in clinical practice. Yet, late-onset and early-onset TGAFs differ fundamentally in pathogenesis, clinical presentation, and management. Temporal classification thus provides a clinically meaningful framework for understanding and managing TGAF. Explicitly distinguishing between these temporal subtypes enables clinicians to formulate pathogenically grounded hypotheses, optimize diagnostic protocols, and implement time-sensitive interventions, thereby reducing the likelihood of diagnostic delay, inappropriate treatment, and preventable complications.

The reasons for diagnostic delay are multifactorial. Its symptoms are nonspecific and overlap with those of gastroesophageal reflux, aspiration, and tumor progression. Clinical suspicion of TGAF declines sharply after the early postoperative period. Consequently, most patients are monitored solely for oncologic recurrence, and their respiratory symptoms are frequently misattributed to metastatic disease or radiation pneumonitis before a fistula is considered, often until recurrent pulmonary infections or pathognomonic signs such as food expectoration appear.

Among the proposed mechanisms, some are well supported while others remain speculative. Peptic ulceration is the most plausible pathway, with endoscopic studies documenting ulcer prevalence of 6.1% to 19.4% (19,20) and caselevel evidence linking ulcer perforation to fistula formation. Radiationinduced injury is radiobiologically plausible but supported mainly by case reports. Tumor recurrence is oncologically established, though its proportion among lateonset TGAF cases is unquantified. In contrast, chronic ischemia and mechanical stress remain largely inferential. This distinction matters: therapies targeting ulcer healing or tumor control are mechanistically grounded, whereas those aimed at ischemia or mechanical stress are more speculative.

The evidence base for management is uneven. Endoscopic stenting has the strongest support, with high immediate technical success and reasonable short-term relief, though longterm outcomes are limited. Surgical repair is supported by case series and systematic reviews confirming feasibility, but its superiority over endoscopy remains unproven. Many clinical decisions rely on singlecenter experience—a limitation that should be explicitly acknowledged given the rarity of the condition.

Long-term survival in late-onset TGAF is primarily determined by the underlying oncologic status and the patient’s physiological reserve. Patients who survive the acute phase often remain dependent on nutritional support, experience recurrent pulmonary infections, and suffer from diminished quality of life due to chronic respiratory impairment. Beyond survival, therefore, longterm nutritional status and pulmonary function represent major concerns. Prospective data on these outcomes are virtually absent, and most reports focus exclusively on shortterm technical success. This represents a critical knowledge gap.

Based on the evidence synthesized in this review, an initial framework for managing late-onset TGAF could be proposed. Given that late-onset fistulas may manifest months or even years after surgery, the current follow-up strategy, focused primarily on tumor surveillance, is inadequate for detecting this complication. Longterm followup should extend beyond the conventional 3- to 5-year window. In patients presenting with recurrent pneumonia, chronic cough, postprandial choking, or progressive weight loss despite the absence of tumor recurrence evidence, late-onset TGAF should be actively considered regardless of symptom onset timing. Initial evaluation should include contrast esophagram with water-soluble contrast and high-resolution chest CT, with attention to the gastric conduit-airway interface. If suggestive, urgent bronchoscopy and gastroscopy should confirm the diagnosis, localize the fistula, and assess tissue viability. Endoscopic stenting is the firstline intervention in most patients, particularly those with poor physiological reserve, active infection, or advanced tumor burden, offering prompt symptom relief and infection control with acceptable risk. Surgical repair should be reserved for patients with adequate physiological reserve, controlled infection, and favorable local tissue conditions, but must be individualized with explicit acknowledgment of substantial perioperative risks.

In summary, late-onset TGAF is a rare, life-threatening, and underrecognized complication. Owing to its rarity and multifactorial pathogenesis, no consensus treatment guideline currently exists for this condition. The available evidence remains fragmented, derived predominantly from case-level observations, and leaves many important clinical questions unanswered. Until prospective registries and collaborative studies provide higher-level evidence, management must be guided by multidisciplinary expertise, careful physiological assessment, and clear communication of prognosis and therapeutic goals.

Supplementary

The article’s supplementary files as

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.