The histochemical findings were consistent with PD, and there were no neoplastic changes in the neighboring organs. Therefore, we diagnosed the patient with primary PD of the esophagus. Esophageal PD is relatively uncommon [2,3,4,5,6,7,8,9,10]. However, fewer cases could be identified as “true” primary esophageal PD, i.e., those that showed evidence of glandular differentiation and excluded intraepithelial metastases of carcinomas arising in other sites [3,4,5,6,7,8] (Table 1).

Two cases of extramammary PD with similar findings (one primary esophageal and one primary scrotal) were reported to be acantholytic anaplastic PD [5, 9]. In these two cases, the tumor cells were relatively well defined. However, in the present case, the tumor cells grew sporadically and were isolated within the pre-existing epithelium without forming conspicuous clusters. There have been no previous reports of EMPD with such characteristics.

In our case, immunohistochemical findings suggested E-cadherin depletion in the tumor cells. Therefore, it is possible that the weakening of E-cadherin was the cause of the reduced connectivity between cells and the isolated and scattered distribution of some tumor cells without the formation of large clusters. Deactivation of Cadherin1 has been hypothesized to be the mechanism underlying the loss or weakening of E-cadherin expression. This theory has been proposed for invasive lobular carcinoma and gastric poorly differentiated adenocarcinoma; however, it is not known whether this mechanism is true for PD [12, 13]. Since PD is also an adenocarcinoma, it is assumed to be caused due to similar reason; however, it is desirable to elucidate the mechanism by accumulating cases.

In this case, the distribution of tumor cells also made it difficult to identify the distribution of the lesion preoperatively. PD is reported as a tumor that is difficult for surgeons to determine the line of resection properly because the boundaries of the lesion are not clear, and the tumor tends to spread unexpectedly beyond macroscopic tumor boundaries [14]. Iodine staining in esophageal endoscopy utilizes the chemical reaction of iodine and glycogen in the squamous cells of the esophageal epithelium. Iodine staining can identify cancer in only a little more than 80% of cases [15]. However, thinning or loss of glycogen-containing non-neoplastic esophageal squamous epithelium may result in reduced iodine staining and may be unstained. To identify the neoplastic lesion, the boundary between the non-neoplastic squamous epithelium and tumor cells must be clear. However, the scattered distribution of tumor cells, as in this case, may have contributed to the difficulty of localization diagnosis. The prognosis for EMPD depends on the surgical margins [1]. This may also be true for primary esophageal PD. Although it is difficult to identify the extent of PD in primary esophageal lesions by iodine staining, this case demonstrated that the neoplastic boundaries can be almost well identified by the vascular changes observed by NBI. The standard treatment of esophageal PD is a surgical procedure similar to that for conventional squamous cell carcinoma [3, 5, 8]. There are high expectations that endoscopic treatment can complete the treatment for PD.