Natural Course of Asymptomatic Upper Gastrointestinal Subepithelial Lesion of 2 cm or Less in Size

There is limited evidence of a natural course of an upper gastrointestinal (UGI)-subepithelial lesion (SEL) of 2 cm or less in size. This study aims to determine the natural course of UGI-SELs and find the risk factors of the endoscopic and endoscopic ultrasonography (EUS) findings associated with an increase in size. The medical records of 2539 patients with UGI-SELs between 2004 and 2016 were reviewed retrospectively. A total of 672 SELs of 2 cm or less in size were analyzed through EUS and followed up for at least 36 months. The mean follow-up duration was 68 months (range: 36–190 months), and 97 SELs (14.4%) showed an increase in size with a mean increase rate of 1.2 mm/year. Initial size (aOR 1.03, 95% confidence interval (CI) 1.01–1.06), an endoscopic finding of a hemorrhagic spot (aOR 3.13, 95% CI 1.14–8.60), and an EUS finding of a lesion in the fourth layer (aOR 1.87, 95% CI (1.21–2.88) were related to an increase in size. An endoscopic finding of translucidity (aOR 0.28, 95% CI (0.10–0.76) and an EUS finding of calcification (aOR 0.30, 95% CI 0.09–0.95) were inversely related to an increase in size. There was no death related to UGI-SELs during the follow-up. While most UGI-SELs of 2 cm or less in size showed no significant size change and favorable prognosis, an individualized follow-up strategy needs to be considered in case of the presence of hemorrhagic spots and lesions in the fourth layer.


Introduction
Subepithelial lesions (SELs) of the upper gastrointestinal (UGI) tract include a variable differential diagnosis, which covers benign and malignant tumors. Endoscopic ultrasonography (EUS) can be beneficial to the differential diagnosis of SELs and decisions on management strategies because it provides useful information regarding the location, size, and echogenic characteristics of tumors [1,2]. Recently, the use of artificial intelligence has been suggested as a less-invasive approach for endoscopic diagnosis including UGI-SELs with better diagnostic accuracy to differentiate potential malignant lesions from other subepithelial lesions [3][4][5][6][7].
In some instances, endoscopists need to decide whether to perform a surgical or endoscopic procedure for removal or EUS-guided intervention to sample tissues, especially in atypical cases or lesions with a malignant potential [8][9][10][11][12][13]. However, resection is not recommended for most UGI-SELs considered to be benign tumors without any clinical symptoms. A serial follow-up can be more appropriate if the size of UGI-SELs seems stable [14][15][16]. According to the European Society of Gastrointestinal Endoscopy guideline, EUS-guided sampling is not recommended for <2 cm SELs located in the esophagus or stomach, and a follow-up is recommended [17]. However, with respect to a follow-up, clear consensus on interval and period is lacking [18][19][20].
Thus far, long-term follow-up data for benign-looking UGI-SELs and studies on the possible effect of an increase in size of SELs on a patient's prognosis are lacking [21,22]. 2 of 9 In the present study, we aimed to evaluate the natural course of a benign-looking UGI-SEL 2 cm or less in size and the risk factors of a baseline endoscopic or EUS finding for predicting the increase in size.

Patients
This study was a retrospective observational study. From January 2004 to December 2016, UGI-SELs were identified in 2539 patients through esophagogastroduodenoscopy (EGD) at Chonnam National University Hospital, Gwangju, Korea. The following patients were included: (1) patients who underwent EUS for the evaluation of UGI-SEL and (2) patients who had serial EGD to check the change in the size of SEL for at least 36 months. The following patients were excluded: (1) patients with >2 cm SELs, (2) patients who underwent surgical or endoscopic resection immediately after initial EUS because of malignant potential, (3) patients who suggested extrinsic compression or no definite SEL in EUS finding, and (4) patients without definite SEL in the follow-up EGD. A total of 672 patients constituted the study population who met the inclusion and exclusion criteria ( Figure 1). Of these patients, 103 had multiple UGI-SELs, and only the largest SEL was analyzed as a representative lesion. Therefore, 672 UGI-SELs were analyzed in 672 patients.

Patients
This study was a retrospective observational study. From January 2004 to 2016, UGI-SELs were identified in 2539 patients through esophagogastroduod (EGD) at Chonnam National University Hospital, Gwangju, Korea. The followin were included: (1) patients who underwent EUS for the evaluation of UGI-SE patients who had serial EGD to check the change in the size of SEL for at least 3 The following patients were excluded: (1) patients with >2 cm SELs, (2) pat underwent surgical or endoscopic resection immediately after initial EUS b malignant potential, (3) patients who suggested extrinsic compression or no de in EUS finding, and (4) patients without definite SEL in the follow-up EGD. A t patients constituted the study population who met the inclusion and exclusi ( Figure 1). Of these patients, 103 had multiple UGI-SELs, and only the larges analyzed as a representative lesion. Therefore, 672 UGI-SELs were analyz patients.
This study was conducted in accordance with the ethical guidelin Declaration of Helsinki. The study protocol was approved by the ethics commi Chonnam National University Hospital (Institutional Review Board Numbe 2020-116; approval date: 5 July 2020).  This study was conducted in accordance with the ethical guidelines of the Declaration of Helsinki. The study protocol was approved by the ethics committee of the Chonnam National University Hospital (Institutional Review Board Number: CNUH-2020-116; approval date: 5 July 2020).

Endoscopic and EUS Evaluation
All examinations were performed by experienced endoscopists (each with experience of >3000 endoscopic examinations). SEL was measured using a visual estimate of the largest diameter of open biopsy forceps during endoscopy as a size criterion (open diameter of 8.0 mm; reusable type, Olympus, Tokyo, Japan). Endoscopic and EUS findings were assessed independently by two endoscopists (DH Kim and SY Cho). Endoscopic and EUS findings were re-evaluated in case of disagreement between the two endoscopists, who reviewed the findings and then reached an agreement. The following endoscopic features were recorded: (a) initial size, (b) presence of umbilication or central dimpling on the surface, (c) presence of erosion or ulcer, (d) presence of erythema, (e) presence of hemorrhagic spot, and (f) presence of translucidity ( Figure 2). EUS probes (UM-2R/UM-3R: Olympus) and a probe-driving unit (MH-MAJ-1720: Olympus) were initially used to map the lesions. The imaging frequency of the probe was 12 or 20 MHz. EUS was performed by experienced endosonographers (DH Kim, SY Park, and JS Rew). Lesions were scanned after the UGI tract was filled with distilled water. The following EUS features were also recorded: (a) maximal and cross-sectional diameters, (b) layer of origin, (c) homogeneity (homogeneous or heterogeneous), (d) distinctness of the border (distinct or indistinct), (e) presence of anechoic foci (duct-like structure), (f) presence of hyperechoic foci with acoustic shadowing (suggesting calcification), (g) deep attenuation, and (h) presence of septation ( Figure 2). SELs were classified as superficial (S-type) and deep (D-type) types that were modified based on the proposed classification proposed by Park et al. [23]. In the S-type, lesions originated in the second and/or third layer. In the D-type, lesions were found in the fourth layer with or without extension into the fifth layer.

Endoscopic and EUS Evaluation
All examinations were performed by experienced endoscopists (each with experience of >3000 endoscopic examinations). SEL was measured using a visual estimate of the largest diameter of open biopsy forceps during endoscopy as a size criterion (open diameter of 8.0 mm; reusable type, Olympus, Tokyo, Japan). Endoscopic and EUS findings were assessed independently by two endoscopists (DH Kim and SY Cho). Endoscopic and EUS findings were re-evaluated in case of disagreement between the two endoscopists, who reviewed the findings and then reached an agreement. The following endoscopic features were recorded: (a) initial size, (b) presence of umbilication or central dimpling on the surface, (c) presence of erosion or ulcer, (d) presence of erythema, (e) presence of hemorrhagic spot, and (f) presence of translucidity ( Figure 2). EUS probes (UM-2R/UM-3R: Olympus) and a probe-driving unit (MH-MAJ-1720: Olympus) were initially used to map the lesions. The imaging frequency of the probe was 12 or 20 MHz. EUS was performed by experienced endosonographers (DH Kim, SY Park, and JS Rew). Lesions were scanned after the UGI tract was filled with distilled water. The following EUS features were also recorded: (a) maximal and cross-sectional diameters, (b) layer of origin, (c) homogeneity (homogeneous or heterogeneous), (d) distinctness of the border (distinct or indistinct), (e) presence of anechoic foci (duct-like structure), (f) presence of hyperechoic foci with acoustic shadowing (suggesting calcification), (g) deep attenuation, and (h) presence of septation ( Figure 2). SELs were classified as superficial (S-type) and deep (D-type) types that were modified based on the proposed classification proposed by Park et al. [23]. In the S-type, lesions originated in the second and/or third layer. In the D-type, lesions were found in the fourth layer with or without extension into the fifth layer.

Endpoints
The primary endpoint was to analyze predictors for an increase in size of UGI-SELs. The secondary endpoint was to identify the mortality rate associated with an increase in size of SEL. An increase in size of UGI-SEL was defined as a 25% increase in the longest diameter of SEL upon endoscopic inspection.

Endpoints
The primary endpoint was to analyze predictors for an increase in size of UGI-SELs. The secondary endpoint was to identify the mortality rate associated with an increase in size of SEL. An increase in size of UGI-SEL was defined as a 25% increase in the longest diameter of SEL upon endoscopic inspection.

Analysis of Survival
Patients' death, date of death, and death-related disease codes were confirmed by requesting data from the Korea National Statistical Agency (http://kostat.go.kr/portal/ eng/index.action, accessed on 12 October 2020). Disease codes were classified in accordance with the Korean Standard Classification of Disease and Cause of Death 7 of the Korea Informative Classification of Diseases [24,25].

Statistical Analysis
Data were statistically analyzed using SPSS version 25.0 (Armonk, NY, USA: IBM Corp). For continuous variables, differences between groups were evaluated using an unpaired t-test. For discrete variables, differences were expressed as counts and percentages and analyzed using a χ 2 test or Fisher's exact between groups as appropriate. The increase in size of UGI-SELs was subjected to Kaplan-Meier analysis, and differences among the groups were assessed via a log-rank test. Statistical differences between groups were analyzed via a pairwise over strata method. Cox proportional hazard regression was performed to analyze the adjusted odds ratios (aORs) as estimates of the increase in size of UGI-SELs. Relevant variables in regression analysis were controlled. All potentially relevant variables were as follows: age, sex, and baseline characteristics of endoscopic and EUS findings. Statistical significance was set at p < 0.05.

Results
A total of 672 patients with UGI-SELs (mean age of 54.6 ± 10.7 years; 278 males, 41.4%) were included in the final analysis. The mean initial diameter of the SELs was 10.7 ± 4.  Table 1 shows the differences in the demographic, endoscopic, and EUS findings between the patients without an increase in size of UGI-SELs and the patients with an increase in size of UGI-SELs. There was a difference in age at the time of diagnosis of UGI-SEL between patients without an increase in size (54.2 ± 10.6 years) and patients with an increase in size (56.8 ± 10.6 years, p = 0.03). An increase in size of UGI-SELs was more frequently observed in the stomach than in the esophagus or duodenum (p < 0.01). Endoscopic findings such as erythema (p < 0.01) and hemorrhagic spots (p = 0.04) significantly differed between the two groups. Translucidity tended to be less frequently observed in patients with an increase in size (p = 0.06). EUS findings such as longitudinal diameter (p < 0.01), cross-sectional diameter (p < 0.01), and D-type (p = 0.01) also significantly differed between the two groups.

Survival Analysis
The result of the Korea National Statistical Agency showed that 16 patients (16/672, 2.4%) died. The median follow-up duration was 120 months (range of 44-193 months). Of the patients with an increase in size of UGI-SELs, 4 patients (4/97, 4.1%) died. Of the patients without an increase in size of UGI-SELs, 14 patients (12/575, 2.1%) died. The most common cause of death was lung cancer (n = 3), followed by pneumonia (n = 2) and trauma (n = 2). The malignancy of the UGI tract was not a cause of death in all patients who died.

Discussion
In our study, patients' age and the size of UGI-SELs at the time of diagnosis was related with an increase in size of UGI-SELs. Endoscopic finding of hemorrhagic spot and EUS findings of the deep type (lesion in the fourth layer) were related with an increase in size, whereas an endoscopic finding of translucidity and EUS findings of anechoic foci and calcification were inversely related with an increase in size of UGI-SELs. There was no UGI-SELs-related death in patients with UGI-SELs of 2 cm less in size.
The size of UGI-SELs increased in 14.4% of patients with small UGI-SELs (<2 cm in size), which were higher than that described in previous similar studies (3-9%) [26][27][28]. These differences were attributed to the longer follow-up period of our study (median 68 months) compared with those in other studies [26][27][28]. The frequency of the increase in the size of UGI-SELs was higher in the stomach (17.7%) than in the esophagus (7.9%) or duodenum (10.0%). This finding was similar to previous results [27,28]. In our study, age at the initial diagnosis was positively correlated with an increase in size of UGI-SELs, which was similar to Song's study [27].
Gross endoscopic findings associated with an increase in size of UGI-SELs included the initial size and hemorrhagic spot. Hemorrhagic spots of SELs may be a stigma of recent bleeding [27,28]. Gross endoscopic findings such as of translucidity were inversely related with an increase in size of UGI-SELs. The translucidity of SELs may indicate benign cystic components of lymphangioma or cyst [13,25,26].
Limited information is available about the characteristic EUS findings associated with the natural course because of the relatively short follow-up duration and the small number of cases [21,22]. Kim et al. [26] reported that any EUS findings were not related to an increase in size of SELs in 476 patients during the follow-up of 30 months. In the present study, EUS findings such as D-type and absence of anechoic foci or calcification were related to an increase in size of UGI-SELs. In EUS, GIST is often a hypoechoic lesion invading the muscle propria (fourth) layer. Benign SELs (such as Brunner's gland hyperplasia, lipoma, cyst, and ectopic pancreas) often originate from the second or third layers [29]. This difference likely explains why the D-type SEL has a higher rate of size increase than the S-type SEL. Calcification is more commonly observed in leiomyoma (6.5-18%) compared with GIST and schwannoma (0-3.5% and 0-3.7%, respectively) [29]. Since leiomyoma has a relatively low rate of increase in size compared with that of GIST or schwannoma, an increase in size may be less frequently observed in the presence of calcification in SELs. Anechoic foci are commonly observed in benign lesions such as ectopic pancreas [30].
Using the databases of the Korea National Statistical Agency, we found 16 (2.4%) patients who died. Among them, no one died because of UGI cancer, suggesting favorable prognosis.
Our study included the large number of patients with long-term follow-up durations among studies regarding the natural course of small UGI-SELs. While most UGI-SELs of 2 cm or less in size showed no significant size change and favorable prognosis, individualized follow-up strategies need to be considered in cases of the presence of hemorrhagic spots and lesions in the fourth layer.
This study had some limitations. First, it was a single-center single-nation study. Second, in this retrospective study, data were not collected prospectively. Therefore, observation errors might have occurred even though we tried to reduce them. Nonetheless, long-term follow-up data of numerous patients were obtained, and they had a high value as EUS was initially performed. Therefore, our study of UGI-SELs of 2 cm or less in size support the current practice of observing, rather than the histological assessment or removal of such lesions. More clear facts related to the initial EUS finding and the change in the size of SELs should be obtained through subsequent studies.

Conclusions
While most UGI-SELs less than 2 cm in size showed no significant size change and favorable prognosis, individualized follow-up strategies need to be considered in cases of the presence of hemorrhagic spots and lesions in the fourth layer.