Patients’ clinical characteristics
All patients were Chinese, dominantly male (male: female =3: 2), with a median age of 65.6 (range, 61-70) years old. The tumor and adjacent non-tumor tissues were isolated from ESGDAC patients who underwent surgical resection at the Department of Thoracic Surgery, The First Affiliated Hospital of Zhengzhou University from January 2018 to October 2020. All the lesions (range, 1.5-3.5 cm) were located at the middle esophagus. The endoscopic appearance showed a submucosal tumor-like lesion. Gastric mucosal erosion and hyperemia were detected in all five cases, and three cases showed a significantly narrowed esophageal. Thoracoscopic radical resection of esophageal carcinoma was adopted in all cases. All patients, except for one case who died because of severe pulmonary infection at 8 days after surgery, were followed up and remained well without recurrence or rapid aggravation of esophageal diseases (Table 2).
Pathological features
ESGDAC was located mainly in muscularis mucosa, and infiltrated into submucosa and muscularis propria without venous or lymphatic invasion, while perineural invasion could be detected in all cases. A number of neoplastic tubules exhibited a bi-layered structure, with inner luminal and outer epithelial layers; others showed small solid, adenoid and irregular cell nests with poorly differentiated adenocarcinoma. The outer epithelial layers had a squamous appearance, whereas the inner luminal layers showed a columnar epithelial cell differentiation. Histologically, the tumor was composed of moderately-to-poorly differentiated adenocarcinoma with marked cellular atypia, whereas the adjacent stratified squamous showed no intraepithelial neoplasia in all cases. Scattered nests of keratinization (1 case) and goblet cell-like mucus cells in the inner layer of the tubules (4 cases) could be observed (Figure 1).
The results of immunohistochemical analysis were available for all cases with ESGDAC. It was shown that the luminal epithelial cells lining the inside of the ducts were positive for CK7 and CAM5.2, and basal cells were positive for p63 and p40. The two layers of cells were negative for S-100, SOX-10, c-Myb, and CD117, which were always positive for salivary gland-type adenocarcinoma. In all the cases, we detected overexpressed p53 protein in both the inner layer of luminal and outer layer of basal epithelial cells, indicating the occurrence of mutation of p53 gene in all the cases. Furthermore, overexpression of HER-2 (c-erbB2) was identified by immunohistochemical assay and no amplification of HER-2 was noted. The proportion of Ki-67-positive cells was high (range, 10-40%). The results of immunohistochemical assay are displayed in Figure 2.
Molecular features
WES was performed to profile the somatic mutational landscape of ESGDAC, comprising primary tumors and adjacent non-tumor tissues from 3 ESGDAC patients, because of degradation of DNA of the other two patients. To the best of our knowledge, this is the first study to perform WES on ESGDAC cases. Table 3 presents the summary of WES coverage statistics for each patient.
Somatic mutation is the biological mechanism underlying the regulation of gene expression, and it plays a significant role in the study of cancer genomics. Somatic mutation can be categories into three groups by size: somatic SNVs, somatic InDels, and somatic CNVs (Xu 2018). Somatic SNVs was detected by muTect (Cibulskis et al. 2013). MuTect run in default setting by taking tumor and normal sample data as input. A total of 796 somatic SNVs were identified, and 705 were nonsynonymous SNVs. Patient 3 had the highest number of somatic SNVs, which was almost twice higher than that of Patient 1 (Figure 3A). Somatic InDels were analyzed by Strelka (Saunders et al. 2012), and 37 somatic InDels were detected. Patient 2 had the highest number of somatic InDels, which was six times higher than that of Patient 1 (Figure 3B). It has been demonstrated that somatic CNVs aroused as a result of preferential selection, favoring cancer development, and the study of chromosomal deletion and amplification has markedly attracted scholars’ attention (Liang et al. 2016). Control-FREEC was used to detect somatic CNVs (Boeva et al. 2012), and GISTIC was employed to infer recurrently amplified or deleted genomic regions. Patient 2 had the highest number of somatic CNVs. In contrast to the somatic SNVs and somatic InDels, the number of somatic CNVs did not significantly change among the three samples (Figure 3C).
In addition to somatic mutations, CPGs, in which deleterious germline variants are associated with the increased risk of cancer, play an important role in the pathogenesis of cancer. Detection of CPGs may enable genetic counseling and cancer surveillance for early detection, and it may lead to revision of cancer treatment protocols (Park et al. 2018). We identified 30 ESGDAC-associated CPGs via analyzing germline variants in ESGDAC cases. The genes, such as ERBB2, BRD3, AKAP9, MKL1, MYH11, COL1A1, ZFHX3, and KMT2D, were present in 2 of 3 tumors, and the remaining of the CPGs were present in 1 of 3 tumors (Figure 4).
Furthermore, we determined the type of mutational signatures for ESGDAC patients. The results revealed mutational signature 1 as the major signature, influencing ESGDAC patients. The mutational signature 1 was the result of an endogenous mutational process and it was associated with small numbers of small insertions and deletions in the majority of tissue. This signature has been found in all cancer types and in the majority of cancer samples. In addition, mutational signature 5, which exhibited transcriptional strand bias for T>C substitutions at ApTpN context and has also been found in all types of cancer and the majority of cancer samples, might be another important mutational signature, affecting ESGDAC patients (Figure 5). The above-mentioned results suggested that different biological mechanisms might be involved in mutagenesis of ESGDAC.
Finally, we compared somatic variations with the known driver genes and detected candidate driver genes from the tumor samples. A total of 12 ESGDAC-associated driver genes (TP53, ACSL6, PABPC4, AFF1, SPEN, etc.) were identified. TP53 mutation was present in all the three tumors, representing the most frequent genomic alteration in this study (Figure 6A). In addition, MuSiC was used to identify SMGs by taking somatic mutations as input and comparing them with background mutation rate (Dees et al. 2012). The CT, FCPT, and LRT were performed on mutant genes, and the gene with Q-value < 0.2 was retained. The results revealed that TP53 mutation was identified in all the three tumors, and it could be a SMG in ESGDAC (Figure 6B).