Endoscopic diagnosis of non-erosive reflux disease

Abstract

Non-erosive reflux disease is the predominant subtype of gastroesophageal reflux disease and significantly impairs patient quality of life. Minimal mucosal changes in the esophagus are difficult to detect using conventional white-light endoscopy; thus, reflux monitoring is required for accurate diagnosis. With advances in endoscopic technology, innovative procedures have recently identified minimal mucosal alterations in the esophagus of non-erosive reflux disease patients. These developments have considerably enhanced diagnostic accuracy and provided a foundation for clinical practice. This mini-review examines advances in common endoscopic techniques. It intends to improve clinical diagnosis and management, and to reduce the economic burden and patient suffering that result from the disease.

Key Words: Gastroesophageal reflux disease; Non-erosive reflux disease; Endoscopy; Diagnosis; Artificial intelligence

Core Tip: This mini-review emphasizes the importance of advanced endoscopic imaging for detecting subtle microstructural changes in the diagnosis of non-erosive reflux disease. We evaluate the diagnostic validity of these findings and propose a practical clinical framework. Looking forward, we highlight the emerging potential of artificial intelligence for automated analysis of imaging data. We also advocate the development of integrative, multimodal diagnostic strategies combining symptom questionnaires, endoscopic methods, functional measures, and clinical biomarkers to enhance non-erosive reflux disease management.

INTRODUCTION

The Montreal Consensus defines gastroesophageal reflux disease (GERD) as a condition in which the reflux of stomach contents into the esophagus causes troublesome symptoms and/or complications[1]. Heartburn and regurgitation are considered the most typical symptoms. Three subtypes of GERD exist: (1) Non-erosive reflux disease (NERD); (2) Reflux esophagitis (RE); and (3) Barrett’s esophagus. The most common manifestation is NERD, accounting for 60%-70% of patients[2].

Patients with NERD exhibit typical reflux-related symptoms but show no evidence of mucosal breaks on endoscopy[1]. Current common auxiliary diagnostic approaches for NERD include symptom questionnaires, such as the Gastroesophageal Reflux Disease Questionnaire (GerdQ) and Reflux Disease Questionnaire, as well as acid-suppressive therapies, such as proton pump inhibitors (PPIs) and potassium-competitive acid blockers[3,4]. However, some patients lack typical reflux symptoms or fail to respond to acid suppression, making a definitive diagnosis challenging based solely on these approaches. The 24-hour multiple intraluminal impedance-pH (MII-pH) monitoring is the gold standard method for diagnosing GERD and is considered a crucial step in the diagnosis of NERD. According to the Lyon Consensus 2.0[5], an acid exposure time (AET) > 6% can be used as an objective criterion for the diagnosis of NERD. This criterion distinguishes NERD from reflux hypersensitivity (RH), where there is normal acid exposure but strong symptom-reflux correlation, and from functional heartburn (FH), where there is normal acid exposure with no symptom-reflux correlation. However, it has limited clinical application owing to various issues such as high cost, long monitoring period, discomfort, day-to-day variability, and nonacceptance[6]. In view of variations in acid reflux exposure, Lyon Consensus 2.0 utilizes 96-hour wireless reflux monitoring as another diagnostic tool for diagnosing GERD[5]. This is limited, however, by its high cost and inability to identify non-acidic refluxes. Thus, finding alternative ways of diagnosing NERD is very important.

Even though there are no visible mucosal changes noted in white-light endoscopy (WLE) in NERD cases, studies on pathology show that there are some changes due to inflammation in the mucosa of the esophagus (Figure 1)[7-9]. The pathological changes include basal cell hyperplasia, inflammatory cells in the tissue, elongated lamina propria papillae, and enlarged spaces between squamous epithelial cells. It can be concluded from the above studies that the esophageal mucosa in NERD is not completely normal. Minimal change lesions (MCLs) in NERD were detected through novel endoscopic techniques at the gastro-esophageal junction. These results coincide with histopathological studies[9], showing mucosal abnormality in NERD. The characteristic MCLs found in NERD patients can serve as intuitive evidence to support the diagnosis, which is helpful in reducing the reliance on invasive reflux monitoring in specific cases and aiding in the proper administration of acid suppressants. This article reviews advances in various endoscopic techniques to improve clinical diagnosis and treatment of NERD.

Figure 1 Histological characteristics of the esophageal epithelium stained with hematoxylin-eosin[9]. A: Mild hyperplasia of basal cells; B: Increased thickness of the basal zone and elongation of papillae in the epithelium, accompanied by neutrophil and eosinophil infiltration. Adapted from Zhang et al[9]. Citation: Zhang NN, Ma YM, Sun Q, Shi LL, Xie Y, Zou XP. Evaluation of Minimal Change Lesions Using Linked Color Imaging in Patients With Nonerosive Reflux Esophagitis. J Clin Gastroenterol 2022; 56: 405-411. Copyright ©Wolters Kluwer Health, Inc. 2021. Published by Wolters Kluwer Health, Inc. (Supplementary material).

ENDOSCOPIC DIAGNOSIS OF NERD

Blue laser imaging

Blue laser imaging (BLI) uses two monochromatic lasers, with wavelengths of 410 and 450 nanometers, to enhance the visualization of the mucosal surface vasculature and microstructures. It offers four observation modes: (1) WLE; (2) Linked color imaging (LCI); (3) BLI; and (4) BLI-bright mode. This technology addresses the limitations of dark-field imaging associated with earlier spectral methods[10]. When combining BLI with magnification endoscopy (ME) (BLI + ME) to examine the esophageal mucosa, NERD patients typically present type II or type A intrapapillary capillary loop (IPCL) changes (Figure 2)[11]. Currently, BLI is mainly applied to detect early-stage cancers. Diao et al[11] demonstrated that BLI + ME provided diagnostic accuracy comparable to magnifying narrow band imaging (NBI) in identifying early esophageal cancer. Additionally, compared with Lugol’s iodine chromoendoscopy, BLI + ME performed better in differentiating malignant from benign lesions and significantly improved concordance between endoscopic and pathological diagnoses. Only two of the 104 type A IPCL lesions were malignant, suggesting high sensitivity (91.9%) and specificity (95.2%) of the type A IPCL in detecting inflammatory lesions. BLI is also useful for diagnosing laryngopharyngeal reflux. Wang et al[12] reported that BLI showed high sensitivity (89.71%) and specificity (73.68%) for diagnosing laryngopharyngeal reflux by detecting brown patches indicating abnormal IPCL changes in the postcricoid region. Furthermore, BLI demonstrated high concordance (Kappa = 0.654, P < 0.001) with the reflux symptom index and reflux finding score. However, the study failed to exclude FH or RH patients due to the absence of 24-hour MII-pH monitoring as the gold standard. As a novel endoscopic technique, BLI improves diagnostic accuracy, provides economic benefits, and enhances pathological consistency. Nonetheless, there are several limitations associated with this diagnostic tool that include the low rate of implementation, the necessity of specific training, and technical problems related to the equipment. Currently, the data that could confirm its effectiveness in the diagnosis of NERD are insufficient since there is a lack of high-quality studies. However, the high efficiency of visualization of mucosal vasculature through the BLI technique makes it possible to identify IPCL changes among individuals with NERD. Further multicenter, large-scale trials are needed to determine its value in NERD diagnosis and management.

Figure 2 Subtle changes of non-erosive reflux disease under blue laser imaging and linked color imaging. A: Increased number, dilation, and tortuosity of intrapapillary capillary loops under blue laser imaging combined with magnifying endoscopy; B: Reddish mucosal changes and prominent gastric mucosal folds observed at the squamocolumnar junction under linked color imaging. BLI: Blue laser imaging; LCI: Linked color imaging.

LCI

LCI employs special color enhancement techniques to provide excellent color contrast within the red spectrum. Consequently, originally white regions appear whiter and brighter, and originally red areas become markedly redder[13]. LCI facilitates rapid screening and identification of minor color alterations in NERD. Under LCI mode, MCLs present as erythema, blurred Z-line, mucosal fragility, reduced vascularity, whitish turbid discoloration, and accentuated edema or mucosal folds (Figure 2)[14]. Deng et al[15] first demonstrated that LCI significantly enhanced detection rates of MCLs in NERD, while also improving interobserver agreement and intraobserver reproducibility. Zhang et al[9] further validated this approach histopathologically, finding that LCI displayed higher sensitivity compared to WLE in detecting NERD-associated MCLs. The research also identified a statistically significant correlation between endoscopic MCLs and biopsy-proven histological abnormalities, suggesting the potential diagnostic utility of LCI for NERD. However, both studies had methodological limitations: Neither excluded FH patients from their cohorts, and MCLs were interpreted subjectively, which included intrinsic diagnostic heterogeneity and potential observer bias. Sun et al[16] pioneered the quantification of mucosal chromatic characteristics under LCI, proposing the R/(G + B) ratio as an objective endoscopic parameter for evaluating gastrointestinal mucosal lesions. This metric had 95.5% concordance with histopathological diagnosis, significantly outperforming WLE (45.5%) and BLI (65.9%) (P < 0.001). However, the diagnostic performance had limitations, with sensitivity and specificity rates of 51.4% and 77.3%, respectively. Future directions include developing multiparametric models integrating real-time image analysis to enhance the clinical utility of LCI in gastrointestinal disease assessment.

NBI

NBI is widely applied in clinical settings as an endoscopic technique that makes use of hemoglobin’s selective absorption of wavelengths which have less dispersion and greater depth of penetration compared to normal white light, thereby allowing structural visibility to be greatly improved. NBI with ME enables a detailed assessment of the alterations such as IPCLs, micro-erosions in the distal esophagus, islands of columnar epithelium, ridge/villous patterns (RVP), and increased vascularity at the squamocolumnar junction (Figure 3)[17]. In the study conducted by Parikh et al[18], increased vascularity, micro-erosions, and irregularity of the pit pattern were found to occur significantly more often among subjects suffering from NERD than in asymptomatic controls when examined with NBI. Of interest is the fact that the combination of increased vascularity and micro-erosion had an outstanding specificity of 98.3%. In an international multicenter randomized control trial[19], it has been found that all three morphological characteristics (i.e., IPCL tortuosity, RVP, and micro-erosions) collectively have high specificity (86%) and moderate sensitivity (60%) when compared to 48-hour pH monitoring test as the gold standard. RVP had high specificity (81%) in correlation with AET, and was associated with improvement following PPI therapy; thus, RVP may be used as an endoscopic biomarker for diagnosis of NERD. However, no correlation has been observed between GerdQ scores and mucosal lesions seen under NBI endoscopy, indicating that mucosal lesions do not correlate with symptom severity. Symptom severity is also linked to psychological coping strategies, visceral sensitivity, and other factors[2]. NBI makes it possible to recognize changes occurring in the mucosa that cannot be detected using WLE, and it shows great specificity for discriminating between normal mucosa, NERD and RE. The early versions of NBI system were hindered by low luminosity and poor distal resolution, thus impairing both clarity and field of vision. These problems have been solved considerably with the increase of luminosity in the new generation of endoscopic devices (such as Olympus EVIS X1). It is important to carry out further studies to include large-scale comparative analysis of NBI against other methods, practical examination of potential biomarkers like RVP, and combination of methods with new techniques.

Figure 3 Minimal lesions of non-erosive reflux disease under narrow band imaging. A: Increased number, dilation, and tortuosity of intrapapillary capillary loops; B: Micro-erosions identified around the Z-line. NBI: Narrow band imaging.

I-scan

I-scan utilizes white light as the illumination source to reconstruct endoscopic images into enhanced virtual images on the basis of the principle of real-time digital image post-processing. I-scan technology includes three separate modes for image enhancement, namely surface enhancement (SE), contrast enhancement, and tone enhancement. Under tone enhancement mode, punctate erythema, triangular lesions with elongated pits, minute erosions, and blurred vessels can be clearly visualized. In a cohort study[20], researchers found that patients who had GERD demonstrated more MCLs than non-GERD patients and healthy controls. The use of SE technology could be valuable for the timely diagnosis of NERD during upper endoscopy for heartburn. However, subsequent research found no significant difference in MCL prevalence by SE between patients with confirmed GERD and those without GERD[21]. Such discrepancies suggest the need for further verification on whether MCLs are correlated with GERD. The HOYA Corporation of Japan has introduced an innovative optical enhancement (OE) technique, which has been integrated into next-generation endoscopic systems[22]. This OE technique provides better image contrast compared to conventional I-scan (Figure 4). According to a prospective study[23], the OE technique exhibited sensitivity, specificity, and accuracy of 94.4%, 61.9%, and 82.4% respectively in diagnosing NERD against 24-hour MII-pH testing. Meanwhile, sensitivity, specificity, and accuracy in relation to histological results stood at 96.5%, 50%, and 73.6% respectively for the OE technique. Furthermore, there were satisfactory agreement scores both for interobserver and intraobserver assessments. Given its high sensitivity and accuracy in detecting abnormal IPCLs, the OE system holds broad prospects for application in the diagnosis of NERD. However, OE investigation could result in prolongation of endoscopy time due to the detailed procedure involved, while its relatively low specificity may lead to high false-positive results. Current research on the clinical value of OE technology in the diagnosis and management of NERD remains limited, likely due to its low adoption rate and the requirement for specialized operator training.

Figure 4 Optical enhancement improves visualization of intrapapillary capillary loops and their morphological changes compared to white-light endoscopy. A: Under white-light endoscopy, intrapapillary capillary loops (IPCLs) are difficult to visualize; B: Optical enhancement technology significantly improves vascular contrast, facilitating clearer visualization of IPCLs; C and D: Optical enhancement demonstrates increased number, dilation, and tortuosity of IPCLs.

Autofluorescence imaging endoscopy

Autofluorescence imaging (AFI) endoscopy takes advantage of the natural fluorescence emitted by the body tissues to differentiate between normal and pathological tissues[24]. Normal esophageal mucosa appears green in AFI, whereas the esophageal mucosa with subtle changes detected by WLE appears pink or purple in AFI[25]. According to Wang et al[26], the sensitivity of the AFI technique was significantly higher compared to WLE (77% vs 21%), while the accuracy of the former also exceeded the accuracy of the latter (67% vs 52%) when using longitudinal purple lines more than 1 cm in the distal esophagus as diagnostic criteria (Figure 5) with 24-hour MII-pH monitoring as the gold standard[27,28]. And the mucosal alteration was correlated with acid reflux. However, AFI showed lower specificity than WLE (53% vs 97%). These findings suggest that AFI can identify subtle mucosal changes undetectable by conventional WLE, which holds significant clinical implications for improving GERD diagnostic rates. However, this study failed to differentiate between GERD subtypes, and its limited specificity may lead to misdiagnosis of some FH patients as NERD cases. A prospective observational study[27] reported that AFI achieved sensitivity (90.5%) and specificity (90.0%) in distinguishing NERD from FH, suggesting its potential as a complementary diagnostic tool. Nevertheless, no statistically significant difference (P = 0.07) between NERD and FH was found, likely due to the small number of participants included, which questions the validity of the results. AFI provides real-time endoscopic assessment, and the diagnostic criteria are relatively straightforward, reducing the complexity of diagnosis. Nevertheless, further research involving large-scale multicenter studies is necessary. In particular, it is important to understand the histopathology behind minimal changes and the use of the technique in diagnosing NERD.

Figure 5 Autofluorescence imaging and confocal laser endomicroscopy reveal distinct features of normal esophageal mucosa and non-erosive reflux disease[27,28]. A: Normal esophageal mucosa under autofluorescence imaging; B: Longitudinal purple lines visualized under autofluorescence imaging; C: Esophageal mucosal appearance in healthy controls under confocal laser endomicroscopy; D: Dilated intercellular spaces and increased, dilated intrapapillary capillary loops (arrow) in non-erosive reflux disease patients under confocal laser endomicroscopy. Descriptions of subfigures A and B (adapted from Luo et al[27]). Citation: Luo X, Guo XX, Wang WF, Peng LH, Yang YS, Uedo N. Autofluorescence imaging endoscopy can distinguish non-erosive reflux disease from functional heartburn: A pilot study. World J Gastroenterol 2016; 22: 3845-3851. Copyright ©The Author(s) 2016. Published by Baishideng Publishing Group Inc.; descriptions of subfigures C and D (adapted from Chu et al[28]). Citation: Chu CL, Zhen YB, Lv GP, Li CQ, Li Z, Qi QQ, Gu XM, Yu T, Zhang TG, Zhou CJ, Rui-Ji, Li YQ. Microalterations of esophagus in patients with non-erosive reflux disease: In-vivo diagnosis by confocal laser endomicroscopy and its relationship with gastroesophageal reflux. Am J Gastroenterol 2012; 107: 864-874. Copyright ©The American College of Gastroenterology 2012. Published by Wolters Kluwer Health, Inc. (Supplementary material).

Confocal laser endomicroscopy

Confocal laser endomicroscopy (CLE) is an emerging endoscopic technique based on confocal laser microscopy. It provides high-resolution images at up to 1000× magnification, enabling histological assessment of living mucosal structures during endoscopy. This technique reduces the need for traditional biopsies or pathological examination and is thus termed “optical biopsy” (Figure 5)[27,28]. Chu et al[28] showed that CLE can accurately identify morphologic changes of IPCLs and dilatation of intercellular spaces (DIS) in the esophagus of NERD patients, verified by 24-hour MII-pH monitoring, compared to healthy subjects. The increased and dilated IPCLs, in conjunction with the presence of DIS, were found to have the highest sensitivity (100%) for NERD diagnosis. These changes were found to be highly correlated with acid reflux episodes. In this regard, CLE is considered a useful tool for diagnosing NERD at a microscopic level. However, attempts to establish quantitative diagnostic criteria using CLE were constrained by a relatively small number of participants and low histopathological positivity rate, which may compromise the persuasiveness of the conclusions. A study using in vivo and ex vivo methods[29] compared patients with RE and those with NERD by detecting abnormal IPCLs in both populations, with the former having higher diameters, numbers, and cross-sectional areas. CLE can quantitatively analyze microstructural patterns of the esophageal mucosa, providing objective markers that correlate well with histopathological findings[28]. Thus, CLE technology has the capacity to serve as a diagnostic method for NERD. Nonetheless, clinical use of CLE is restricted due to its narrow visual field, long procedure time, and high technical demands on endoscopists. Further research should determine the diagnostic cut-off points for NERD and confirm these results using multicenter trials involving more subjects.

APPLICATIONS OF ARTIFICIAL INTELLIGENCE IN GERD

Artificial intelligence (AI) applications in GERD endoscopic diagnosis should be tailored according to disease subtype. AI has primarily been developed to automatically classify and interpret the Los Angeles (LA) grading system for RE, improving accuracy among inexperienced endoscopists. Several studies have successfully developed and refined LA grading models[30,31]. NERD can be diagnosed based on microscopic changes, such as those involving IPCLs, micro-erosions, and RVP. At the current moment, there is no available AI tool capable of identifying these microscopic changes associated with NERD because of the absence of universal diagnostic criteria for NERD. It becomes clear that using only one method for diagnosing and treating NERD is not efficient as our understanding of the disease grows. In the future, more focus needs to be put on multimodal diagnostics taking into account the clinical and endoscopic features as well as physiological functions (such as the results from pH monitoring). Multimodal diagnostics will enable the prediction of treatment response, improve the accuracy of the differential diagnosis among NERD, RH, and FH, and ultimately lead to individualized therapy.

Several considerations are crucial for clinical translation of AI systems[32,33]. First, standardized image processing is crucial. The various endoscopes use different imaging techniques; hence, their resolution and contrasts differ. Therefore, images must be processed in a standardized manner, which entails adjusting colors, reducing noise, and ensuring resolution consistency. Additionally, a common framework that defines minimal lesions will enhance standardized depiction. Second, a decision on annotation strategies and standards is imperative. Since minimal lesions are subjective, it is important to achieve consensus through annotations from experts independently. The validation criteria need to encompass multiple benchmarks such as pH measurement, histological assessments, and reactions to treatment. Third, model validation is important. Current research mainly uses data collected at one center using one machine, with little to no validation conducted outside of that setting. There is a need to validate the use of the technology on different machines among varied patient populations and different disease manifestations to prove its clinical applicability. Finally, other barriers to clinical translation include the need to make real-time diagnoses and to provide clinical utility in prospective randomized controlled trials. Only by effectively addressing these challenges can the NERD AI diagnostic system advance toward clinical application, enabling precise disease diagnosis and treatment.

DIAGNOSTIC CHALLENGES AND FUTURE DIRECTIONS

NERD is a condition in which patients experience reflux symptoms but show no visible mucosal damage on endoscopy. Under WLE, typical findings are lacking, which makes definitive diagnosis difficult. GerdQ fails to effectively distinguish NERD from RE and has low sensitivity when typical symptoms are absent. Acid-suppressive trials are widely used, but some patients fail to respond. Reflux monitoring is the gold standard for diagnosing GERD, but has limited clinical application due to high cost and patient discomfort. Novel endoscopic technologies derived from traditional WLE clearly visualize subtle esophageal mucosal changes, which enhances diagnostic accuracy, reduces patient discomfort, and demonstrates significant developmental potential.

To understand the clinical significance of MCLs, it is essential to clarify their underlying pathological basis. According to the international consensus established by the Esohisto project[34], histologic changes related to GERD primarily involve basal cell hyperplasia, inflammatory cell infiltration, elongation of lamina propria papillae, and DIS in squamous epithelial cells. Multiple studies have confirmed a significant correlation between these histological changes and the MCLs observed endoscopically. For instance, Chu et al[28] employed CLE for in vivo measurements, revealing that NERD patients exhibited significantly higher numbers, diameters of IPCLs and larger intercellular spaces compared to controls. Moreover, CLE-measured intercellular spaces showed high correlation with transmission electron microscopy findings (r = 0.75, P < 0.001), providing direct evidence for the correspondence between MCLs and pathological alterations in NERD patients. Zhang et al[9] found that patients with MCLs positivity under LCI had significantly higher histological scores than those with negativity (4.59 ± 0.32 vs 2.36 ± 0.34, P < 0.01). Desai et al[19] further confirmed that the combination of IPCL tortuosity, RVP, and micro-erosion under NBI was significantly correlated with papillary length (r = 0.336), basal cell thickness (r = 0.347), and intraepithelial eosinophils (r = 0.329). These studies suggest that there is pathologic abnormality in NERD. However, some researchers found conflicting results between endoscopic appearance and histopathology. A prospective study[35] found no significant association between NBI endoscopic features and GERD histology, suggesting histopathology holds greater diagnostic value than NBI endoscopy. There are possible reasons for these findings, including mismatch between biopsy location and endoscopic findings, individual variability, and the inclusion of FH and RH cases. Moreover, there have been no studies showing correlation between particular endoscopic minimal lesions and histopathology.

Although MCLs show significant correlation with histopathology, interpreting these alterations requires consideration of the diagnostic complexity arising from the clinical similarity among NERD, RH, and FH. NERD involves pathological acid reflux, RH symptoms correlate with reflux events but exhibit normal acid exposure, while FH symptoms are unrelated to reflux events. Deng et al’s study[15] defined PPI trial-positive patients as NERD, which does not entirely exclude FH and RH patients, thereby influencing the assessment of diagnostic efficacy. For RH patients, the diagnostic complexity comes from the fact that they experience reflux events, so they might have some changes visible during endoscopy, which would falsely increase the sensitivity rate. FH patients typically display no endoscopic alterations; their inclusion in NERD groups reduces sensitivity, while their placement in control groups reduces specificity. In Zhang et al’s study[9], 39.5% of controls exhibited minimal endoscopic lesions, likely reflecting the inclusion of asymptomatic RH or NERD patients without pH monitoring. In contrast, Luo et al[27] confirmed that strict differentiation between NERD and FH by using pH monitoring increases the diagnostic efficacy of AFI. The overlap of diagnoses signifies that endoscopic examinations cannot be used as a substitute for pH monitoring but should serve as a complementary diagnostic tool. Core endoscopic findings like micro-erosions and RVP, indicate NERD and warrant initiating acid suppression therapy. Conversely, negative endoscopy should prompt pH monitoring to further differentiate FH or RH.

Currently, there is no standardized procedure for diagnosing NERD through endoscopy. Further studies need to be conducted to incorporate recent advancements into existing diagnostic procedures. The Lyon Consensus 2.0 and the ACG Guidelines recommend a hierarchical approach to diagnosing GERD based on the concepts of “unproven GERD” and “proven GERD”, which provides a basis for determining the clinical relevance of minimal endoscopic abnormalities[5,6]. Guidelines define LA-B grade or higher esophagitis under WLE as definitive evidence, with histopathological alterations and hiatal hernia classified as supplementary evidence. Although minimal endoscopic lesions in NERD patients are not included in current diagnostic criteria, the consistency between MCLs and histopathological changes indicates potential auxiliary diagnostic value. The clinical significance of these microscopic alterations may lie in optimizing management pathways for patients with “unproven GERD”. When pH monitoring results fall into borderline ranges (e.g., AET 4%-6%), presence of MCLs supports pathological reflux, guiding clinical decisions. Conversely, negative MCL findings necessitate cautious reliance on pH-impedance monitoring to exclude RH or FH. Incorporating these considerations, Figure 6 proposes, for the first time, a simplified conceptual framework integrating MCLs into GERD diagnosis and management pathways, offering reference for future researches.

Figure 6 Proposed diagnostic algorithm incorporating advanced endoscopy into the diagnostic pathway for gastroesophageal reflux disease and non-erosive reflux disease. AET: Acid exposure time; AFI: Autofluorescence imaging; BLI: Blue laser imaging; CLE: Confocal laser endomicroscopy; GERD: Gastroesophageal reflux disease; LA: Los Angeles; LCI: Linked color imaging; MCLs: Minimal change lesions; MNBI: Mean nocturnal baseline impedance; NBI: Narrow-band imaging; OE: Optical enhancement; PPIs: Proton pump inhibitors.

There are some factors that need to be considered when employing endoscopy techniques in the diagnosis and treatment of NERD patients. Endoscopic timing is crucial. According to current guidelines, PPI therapies should be discontinued at least two weeks before endoscopy examination. This follows the recommendations put forward by the Lyon Consensus 2.0[5,19]. Endoscopy remains an adjunctive diagnostic tool and cannot replace pH monitoring. Detection of typical microscopic lesions such as RVP during endoscopy significantly increases the likelihood of NERD diagnosis, justifying initiation of acid suppression therapy[19]. Conversely, the lack of microscopic lesions requires pH testing to differentiate NERD from FH and RH. Several studies indicate that some endoscopic findings, including RVP and micro-erosions, might resolve after PPI treatment[19]. Thus, these endoscopic findings might be used as markers of mucosal healing and treatment outcome predictors. However, the available evidence is scarce and mainly comes from short-term randomized controlled trials. Future investigations need to examine how the treatment responses to acid suppression therapy vary at different time periods for each endoscopic finding. It will also be necessary to study whether endoscopic findings could be used to predict the treatment outcomes for different therapies.

Although there are clear principles for implementation, certain difficulties still exist, primarily because of considerable variation in defining the minimal lesion under various endoscopic methods. This poses a problem when trying to set a standard diagnosis for NERD. For example, research involving BLI defines the changes in blood vessels as IPCL Type II changes[11], whereas NBI endoscopy defines the same as dilation and tortuosity of IPCLs[19]. These descriptions essentially represent the same reflux-induced inflammatory vascular alterations. Each imaging modality places more importance on specific aspects of these changes: LCI focuses mainly on mucosal discolorations, while NBI and BLI emphasize microvascular modifications. However, the proposed classification scheme serves as a preliminary approach toward describing microscopic lesions to help endoscopists understand MCLs better (Table 1). The classification system groups endoscopic signs according to microvascular or mucosal changes, and then subdivides each group into core features, which exhibit strong correlation with pathological reflux and have high sensitivity and specificity, and supporting features, which suggest pathological reflux, but need corroboration. Considering the current paucity of scientific literature and inconsistency in research results, the classification serves only as a reference that should be validated in the future using multicenter prospective studies.

Table 1 The detailed characteristics of minimal change lesions in non-erosive reflux disease.

Endoscopic change type	Category	Endoscopic feature	Detailed characteristics
Microvascular alteration	Core feature	IPCL abnormalities	Increased number, dilation, and tortuosity of IPCLs
	Supportive feature	Increased vascularity at the SCJ	Dilated, dark-brown capillaries immediately below the SCJ
Mucosal alteration	Core feature	Ridge/villous pattern	Uniform, longitudinally aligned ridges alternating with a villiform pattern
		Microerosions	Mucosal breaks visible only with image-enhanced endoscopy
	Supportive feature	Islands of columnar epithelium	Dark-brown columnar islands surrounded by white squamous mucosa
		Reduced vascularity	Heterogeneous vascular pattern with areas of decreased vascularity
		Blurred Z-line	Indefinite Z-line border
		Accentuated edema or mucosal folds	Prominent gastric-type folds at the SCJ

IPCL: Intrapapillary capillary loop; SCJ: Squamocolumar junction.

Currently, investigations on microlesions utilize mainly subjective diagnoses made by endoscopists, and thus interobserver variability is an unavoidable challenge[15]. Thus, quantitative assessment methods to enhance diagnostic objectivity and reproducibility are necessary. For instance, Sun et al[16] proposed the R/(G + B) ratio to quantify the color characteristics of mucosa under LCI, demonstrating strong consistency with pathological diagnoses. Chu et al[28] utilized CLE to quantify the number, diameter, and intercellular gap distance of IPCLs, providing the most direct quantitative indicators for NERD diagnosis. Current research predominantly focuses on single parameters. Establishing multi-parameter fusion quantitative diagnostic models can significantly improve diagnostic efficacy and support future development of AI-assisted diagnostic systems.

Although several studies show its diagnostic potential, the ACG clinical guideline has not yet recognized endoscopic techniques as a standard diagnostic tool for NERD[6]. Further validation research and unified diagnosis criteria are required.

CONCLUSION

Establishing a multicenter, large-sample prospective study is necessary to validate the NERD microscopic diagnosis system based on enhanced endoscopy. Combining endoscopic techniques with other technologies will improve the diagnostic accuracy, while developing and validating quantitative metrics for endoscopic features. Endoscopic techniques should be upgraded concurrently to allow for clearer and more comprehensive observation of the esophageal mucosa, thereby identifying true NERD patients and providing a more objective basis for NERD treatment. The advancement of endoscopic technologies will facilitate the future development of AI, while the integration of AI with endoscopic techniques can significantly enhance the endoscopic diagnosis and therapeutic efficacy evaluation of NERD. This synergy is expected to reduce the economic burden of the disease, alleviate patient discomfort, and improve diagnostic efficiency for physicians.

ACKNOWLEDGEMENTS

We extend our heartfelt thanks to all the staff of the Department of Gastroenterology at the Second Affiliated Hospital of Xi’an Jiaotong University for their support and contributions to this research.