We retrospectively analyzed patients with suspected SESCC who underwent examination, including both ME and EUS, and then underwent surgery or ER at Sun Yat-Sen University Cancer Center between January 2018 and December 2021. We included patients with suspected SESCC following white light imaging (WLI) screening or other modalities. All patients were pathologically diagnosed with atypical esophageal hyperplasia or SESCC. We excluded patients who received chemotherapy or radiotherapy as an initial treatment after diagnosis and those who were suspected of having lymph node or organ metastases by imaging. The institutional review board of Sun Yat-Sen University Cancer Center approved this study.

Resected complete specimens obtained during surgery or ER were processed and diagnosed by our Center’s pathology department. According to the Paris Endoscopic Classification of Superficial Neoplastic Lesions[16] and the 11^th Edition of the Japanese Classification of Esophageal Cancers[3], in the esophageal mucosa (T1a), lesion involvement included the epithelium (EP) (including high-grade intraepithelial neoplasia (HGIN) and carcinoma in situ), the lamina propria mucosa (LPM), and the muscularis mucosa (MM). Submucosal (SM, T1b) lesions were divided into SM1, SM2, and SM3. These lesion layers featured equivalent thickness and were ordered from shallower (SM1) to deeper (SM3). Since the submucosal thickness remained unknown in endoscopically resected specimens, lesions involving the submucosa to 200 µm or less from the MM were classified as T1b-SM1. Those deeper than 200 µm were considered T1b-SM2/SM3. Thus, in our study, lesion invasion depths were categorized pathologically as pEP/LPM, pMM/SM1, and pSM2/SM3.

The examination procedure was identical to that used in our daily practice. All lesions included were initially examined by conventional endoscopy with WLI. Suspicious lesions were further assessed using magnifying endoscopy with narrow-band or blue laser imaging (ME-NBI/BLI) using a GIF-H260Z (Olympus Corporation, Tokyo, Japan) or EG-L590ZW gastroscope (Fujifilm Corporation, Tokyo, Japan). EUS followed, utilizing 7.5MHz, 10MHz, or 12MHz radical scanning probes (SU 9000, EG-530UR2, Fujifilm; EU-ME2, Olympus) or a 20-MHz miniature probe (UM-DP20-25R, Olympus). Six certified and experienced endoscopists at our center performed all these examinations. The involved endoscopists were divided into junior and senior groups according to their seniority. The senior endoscopist was defined as having a title of Associate Professor or higher with at least 12 years of experience in endoscopy. The junior endoscopist is defined as having a title of Attending Physician or above, with more than 6 years of experience in endoscopy. Residents and trainees did not participate in this study. Each patient's ME-NBI/BLI and EUS were conducted on the same day. The endoscopic findings were later extracted from the electronic medical record.

ME, combined with image-enhanced endoscopy, NBI, or BLI, allows visualization of micro-vessels on the esophageal surface. Intra-papillary capillary loops (IPCL) are basic microvasculature units on the squamous mucosal surface. IPCL forms are used to characterize lesions and predict invasion depth for SESCC. We applied the Japan Esophageal Society (JES) classification scheme, which integrates previous Inoue and Arima classification schemes, presently in widespread clinical use[7,17]. Here, micro-vessels observed by ME were divided into type A and type B. Type A vessels were non-cancerous lesions; type B vessels were abnormal micro-vessels characterized by dilatation, meandering, caliber change, and uneven morphology. These abnormal features were suggestive of cancerous lesions and include three subtypes: B1 (vessels with loop-like formations), B2 (without loops but appearing stretched and markedly elongated), and B3 (highly dilated vessels with calibers more than three times those of B2 vessels). To predict invasion depth, type B1, B2, and B3 vessels corresponded with depths of EP/LPM, MM/SM1, and SM2/SM3, respectively. The subclassification of type B vessels was based upon the indication for ER: Lesions with B1 were absolutely indicated, B2 vessels were relatively indicated, and B3 vessels were contraindicated.

During EUS, a cross-sectional image of the esophageal wall structure was obtained and divided into five layers using a 7.5 MHz radical conventional probe[18]. When using a high-frequency (≥ 20 mHz) miniature probe, the canal wall was depicted as a nine-layer structure if the distance between the probe and mucosa was appropriate. Specifically, the mucosa and submucosa were sonographically divided into an additional four layers. The first and second layers corresponded to the EP/LPM, the third layer to the MM, and the fourth layer to the SM. Specifically, lesions confined to the first and second layers were categorized as EP/LPM; lesions involving the third layer were MM/SM1; lesions that invaded the fourth layer were SM2/SM3. Esophageal cancer usually appears as a hypoechoic lesion that disrupts the normal structure of the esophageal wall, forming images with defects, irregularities, and interruptions.

The diagnostic efficiencies of EUS and ME-NBI/BLI for determining exact invasion depth were evaluated by sensitivity, specificity, and accuracy. A paired χ² test (McNamar's) was used to assess their differences. P values < 0.05 were considered statistically significant. We applied Cohen's kappa to evaluate the consistency of EUS and ME-NBI/BLI with the final pathological result for determining the depth of tumor infiltration[19,20]. The accuracy of ME-NBI/BLI or EUS concerning the clinicopathologic features was assessed using the χ² test or Fisher's exact test. Multivariate logistic regression analysis was performed to identify variables that significantly influenced the performance of ME-NBI/BLI or EUS. SPSS version 25 for Windows software (IBM Inc, Armonk, United States) was used for statistical analyses.

Of the 146 patients who met our enrollment criterion, two were excluded from the analyses. One was because of hemorrhage during ER, which was later converted to surgical resection; this resulted in an incomplete pathological specimen. Another patient was excluded because we could not obtain a clear view during ME-NBI, preventing micro-vessel characterization.

Ultimately 152 lesions in 144 patients were included in this study. Of these, 108 were male (75%), and 36 were female (25%), with a mean age of 61.3 ± 7.5 years. Most tumors were located in the middle thoracic esophagus (82/152, 53.9%), and the main macroscopic type was flat (90/152, 59.2%). The mean tumor size was 22.9 mm (range 5-60 mm). The average time interval between examinations and resection treatment was 18 d (1-82 d). As for treatment selection, 71 lesions were treated by ER, and 81 were treated by surgery. Pathologically, 78 lesions (51.3%) were diagnosed as pEP/LPM lesions, 28 (22.4%) as pMM/SM1, and 46 (30.3%) as pT1b-SM2/SM3. Detailed clinicopathological features of the patients and lesions are shown in Table 1.

The relationships between ME-NBI/BLI or EUS diagnosis and the final pathological result after treatment are listed in Table 2 and Figures 1-3. The overall accuracy of ME-NBI/BLI, based upon the JES classification for determining invasion depth, was 73.0% (111/152), moderately consistent with the pathological results (kappa = 0.58, 95% confidence interval [CI]: 0.48-0.68, P < 0.01). The overall accuracy of EUS for determining invasion depth was 66.4% (101/152), also moderately consistent with the pathological results (kappa = 0.46, CI: 0.34-0.57, P < 0.01).

Open in New Tab   Full Size Figure   Download Figure

Figure 1 A typical case of carcinoma in situ. A: ME-BLI image shows micro-vessels with a loop-like formation (type B1); B: Ultrasonography image shows hypoechoic thickening confined to the first two layers; C: Hematoxylin-eosin staining (× 40) of an endoscopic resection specimen shows that the squamous cell carcinoma is limited to the epithelium, without invasion.

Open in New Tab   Full Size Figure   Download Figure

Figure 2 A typical muscularis mucosal lesion. A: ME-BLI image shows type B2 vessels without loop-like formations but with a stretched and markedly elongated transformation; B: Ultrasonography image shows a hypoechoic lesion invading the third layer with continuous submucosa; C: Hematoxylin-eosin staining (× 40) of a surgical specimen shows a moderately differentiated squamous cell carcinoma invading the muscularis mucosa.

Open in New Tab   Full Size Figure   Download Figure

Figure 3 A typical submucosal lesion. A: ME-NBI image shows micro-vessels dilated more than three times that of B2 vessels (type B3); B: Ultrasonography image shows a hypoechoic lesion invading the fourth layer; C: Hematoxylin-eosin staining (× 20) of a surgical specimen shows a moderately differentiated squamous cell carcinoma infiltrated to the middle third of the submucosa without muscularis propria involvement.

We also compared the diagnostic efficiency of ME-NBI/BLI and EUS for determining the invasion layer according to the indication for ER (Table 3). There was no significant difference in overall accuracy between ME-NBI/BLI and EUS (73.0% vs 66.4%, P = 0.24). For pEP/LPM lesions, ME-NBI/BLI had a higher sensitivity, specificity, and accuracy than EUS (sensitivity 84.6% vs 73.1%; specificity 91.9% vs 81.1%; accuracy 88.2% vs 77.0%), with a significant difference in accuracy (P < 0.01). For pMM/SM1 lesions, ME-NBI/BLI was more sensitive, and EUS had a better specificity (sensitivity 92.9% vs 35.7%; specificity 73.4% vs 91.1%; P < 0.01 for both); the two techniques demonstrated equivalent accuracy (77.0% vs 80.9%, P = 0.51). For pSM2/SM3, ME-NBI/BLI was more specific and EUS was more sensitive (sensitivity 41.3% vs 73.9%, P < 0.01; specificity 98.1% vs 75.4%, P < 0.01); the techniques had equivalent accuracy (80.9% vs 75.0%, P = 0.22). Lastly, of the 41 lesions (41/152, 27.0%) misdiagnosed by ME-NBI/BLI, 24 were corrected by EUS (24/41, 58.5%).

For ME-NBI/BLI, diagnostic accuracy did not vary significantly according to the tumor location, macroscopic type, circumferential occupation, tumor size, or endoscopist grade (Table 4). The accuracy of ME-NBI/BLI increased significantly for HGIN or carcinoma in situ subgroups (P = 0.03). During the multivariate analysis, HGIN and carcinoma in situ were independent risk factors for the accuracy of tumor invasion depth, as assessed by ME-NBI/BLI (odds ratio [OR] = 3.62, 95%CI: 1.43-9.16, P = 0.007).

As for EUS, the overall diagnostic accuracy did not vary significantly according to the tumor location, macroscopic type, differentiation degree, and endoscopist grade (Table 4). Increased circumferential occupation and tumors larger than 3 cm were mostly associated with decreased accuracy (P = 0.06 and P = 0.05, respectively). Using a miniature probe instead of conventional EUS improved accuracy (82.3% vs 49.3%, P < 0.01). In the multivariate analysis, less than a quarter of circumferential occupation and application of a miniature probe were independent risk factors for the accuracy of tumor invasion depth, as assessed by EUS (< 1/4 circumferential occupation: OR = 3.07, 95%CI: 1.04-9.10; application of a miniature probe: OR = 5.28, 95%CI: 2.41-11.59, P < 0.01).

Early-stage detection and treatment of esophageal carcinoma can typically optimize prognosis. Compared with traditional surgery, endoscopic resection is a less invasive and potentially curative treatment for early-stage esophageal cancer. Identification of patients that are candidates for endoscopic resection is crucial. Endoscopic ultrasonography (EUS) and magnifying endoscopy (ME) reliably determine indications for endoscopic resection in patients with superficial esophageal squamous cell carcinoma (SESCC). ME is a widely accepted method for predicting the invasion depth of superficial esophageal cancer with satisfying accuracy. However, the addition of EUS is controversial.

To evaluate the diagnostic efficiency of ME vs EUS for invasion depth prediction, and investigate the influencing factors.

To determine the most suitable preoperative diagnostic modality for Chinese patients with SESCC.

We retrospectively analyzed patients with suspected SESCC who completed both ME and EUS and then underwent endoscopic or surgical resection at Sun Yat-Sen University Cancer Center between January 2018 and December 2021. We evaluated and compared the diagnostic efficiency of EUS and ME according to histological results, and investigated the influencing factors.

EUS and ME demonstrated comparable accuracy for determining the depth of invasion of early-stage esophageal cancers, and EUS can compensate for deficiencies inherent to NBI in some cases. The miniature probe was best suited for detecting early-stage lesions

Preoperative diagnosis of SESCC should be conducted endoscopically using white light and magnification. In China, EUS can be added after obtaining patient consent. Use of a high-frequency miniature probe or miniature probe combined with conventional EUS is preferable.

Future studies are required to explore how to combine the findings of ME and EUS to make a comprehensive preoperative evaluation, instead of solely depending on the experience of endoscopists.