This was a single-centre, retrospective study. This study was approved by ethics committee of Nagasaki University Hospital.

Fifteen consecutive cases of esophageal achalasia which underwent POEM in our institution between August 2010 and January 2012 were enrolled in this study. A clinical summary of the 15 patients (3 males, 12 females; mean age 51.7 years, age range 26–84 years) with esophageal achalasia is shown in Table 1. The dilation grade of the disease was II in 6 patients and III in 1 patient. Prior endoscopic balloon dilation was performed in 5 patients. Two patients with sigmoid type achalasia were included.

Written informed consent was obtained from all subjects, a legal surrogate, the parents or legal guardians for minor subjects prior to their inclusion, in accordance with the Helsinki Declaration and under approval by the Nagasaki University Ethics Committee. All the subjects also consented for the publication of individual clinical details with a written informed consent.

To evaluate improvement of dysphagia objectively, a dysphagia symptom score and the Eckardt score (representative subjective symptom score) were used 11 . The dysphagia symptom score was defined as follows: the worst degree of dysphagia just before the treatment was scored as 10, and the best condition, before the patient developed achalasia, was scored as zero, in accordance with Inoue’s report 10 . The post-procedural degree of dysphagia was scored between 0 and 10 in each patient. Scoring was done before POEM, after POEM on the day of discharge from hospital, and 3, 6, and 12 months later.

The degree and shape of achalasia were assessed by barium passage. The maximum diameter of the thoracic esophagus was measured, and achalasia was classified into grade I (diameter <3.5 cm), grade II (3.5 cm < diameter < 6.0 cm), and grade III (diameter ≥6.0 cm), in accordance with the descriptive rules (Descriptive Rules for Achalasia of the Esophagus, Japan Society of Esophageal Diseases, June 2012, 4th Edition). On the day following POEM, a barium swallow was done to confirm passage of contrast media through the gastroesophageal junction (GEJ) without leakage. Achalasia shapes include straight, flask, and sigmoid types.

Esophageal pull-through manometry was done in all cases before and after the procedure using the Polygraf ID (Medtronic Functional Diagnostics, Denmark). According to the manufacturer’s data, the normal range for LES pressure with this device is from 14.3 to 34.5 mmHg. The manometry study could not be completed in case 15 before POEM because the measuring catheter did not pass through the narrow segment of the esophagus.

The POEM procedure received approval from the institutional review board (IRB) of Nagasaki University Hospital (approval number 10052824, issued on May, 22 2010). Written informed consent was obtained from all patients.

A forward-viewing endoscope with an outer diameter of 9.8 mm (GIF-H260; Olympus, Tokyo, Japan), which is routinely used for upper gastrointestinal examination, was used with a transparent distal cap attachment (MH-588, Olympus). A triangle-tip knife (KD- 640 L; Olympus) was used to dissect the submucosal layer and to divide circular muscle bundles. A coagulating forceps (Coagrasper, FD-411QR; Olympus) was used to coagulate larger vessels prior to dissection and for hemostasis. The VIO 300D (ERBE, Tubingen, Germany) electrogenerator was used, and for final closure of the mucosal entry site, hemostatic clips (EZ-CLIP, HX- 110QR; Olympus) were applied. The procedures were performed under general anesthesia with positive pressure ventilation. The UCR CO₂ insufflator (Olympus) was used with a regular insufflating tube (MAJ-1742; Olympus), which maintained CO₂ insufflation at a constant rate of 1.2 L/min.

The POEM procedures were done as previously described with slight modifications as follows. Submucosal injection was done first at the level of the middle thoracic esophagus, approximately 13 cm proximal to the GEJ. About 10 mL of saline supplemented with 0.3% indigocarmine were injected. A 2-cm, longitudinal mucosal incision was made on the mucosal surface to create a mucosal entry to the submucosal space (dry cut mode, 50 W, effect 3). Then, a technique similar to endoscopic submucosal dissection (ESD) was used to create a submucosal tunnel downwards, passing through the GEJ, and about 3 cm into the proximal stomach (Figure 1a). Spray coagulation mode (60 W, effect 2) was used to dissect the submucosal layer. Larger vessels in the submucosa were coagulated using the forceps in the soft coagulation mode (80 W, effect 5). Dissection of the circular muscle bundle was started at 2 cm distal from the mucosal entry. The sharp tip of the knife was used to capture only circular muscle bundles and then lift up toward the tunnel lumen (Figure 1b). The captured circular muscle bundle was cut by spray coagulation current (60 W, effect 2). Division of the sphincter muscle was continued from the proximal side towards the stomach until it passed 2 or 3 cm distal from the LES. After completion of the myotomy, smooth passage of the endoscope through the GEJ with minimal resistance was confirmed. The incised mucosal entry was closed with about five hemostatic clips after prophylactic antibiotic solution was sprayed inside the tunnel. Successful closure of the mucosal entry was confirmed by the endoscopic appearance. At the end of the procedure, the endoscope was again inserted into the original esophageal lumen down to the stomach, to confirm smooth passage through the GEJ.

A prototype endocytoscope with an integrated dual charge-coupled device (CCD) (Figure 2) (GIF-Y0001, Olympus) was used. The dual-CCD integrated type endoscope has both conventional magnification (80-fold) and ultra-high magnification (450-fold) capabilities, which can be easily interchanged at the push of a button on the endoscope. The clinical use of the prototype endoscope was approved by the hospital’s ethics committee (IRB approval number 10102236, Oct. 29 2010), and written informed consent was obtained from all patients. A mixture of 0.05% crystal violet and 0.1% methylene blue was used during endocytoscopy to obtain images similar to conventional pathological images using hematoxylin and eosin staining. Crystal violet effectively dyes the cytoplasm, whereas methylene blue allows detailed identification of the cell structure, including the nuclei and cytoplasm. Dilution of methylene blue with crystal violet seems to improve visualization of the lesion and may also decrease the potential risk of DNA damage due to a higher concentration of methylene blue 12 . Patients underwent both conventional endoscopy and endocytoscopy at the same time under conscious sedation with intravenous pethidine hydrochloride (35 mg; Opystan, Mitsubishi Tanabe Pharma, Osaka, Japan), supplemented with diazepam (5–10 mg, Takeda Pharmaceutical, Osaka, Japan). After conventional endoscopic observation, ultra-high magnification was applied. The mucosa on the posterior wall 25 cm distal from the incisor was targeted. Still images were obtained from the targeted area, and 3 images of sufficient quality were selected for evaluation. Then, the average nuclei counts of the 3 images before and 3 months after POEM were compared. After endocytoscopic observation, biopsies were taken from the targeted area. All endocytoscopy examinations were recorded, and the location photographed during endocytoscopy coincided with the histopathological images. Nevertheless, complete correspondence between endocytoscopic images and microscopic images is still limited 13 .

Four-mm tissue sections of the formalin-fixed, paraffin wax-embedded samples were sliced and mounted on silane-coated glass slides, dried, deparaffinized with xylene and rehydrated through graded ethanol. Antigen retrieval was performed by boiling for 15 minutes in 10 mM monocitric acid buffer (pH 6.0) for anti-Ki-67 staining and in 10 mM Tris/ethylenediaminetetraacetic acid buffer (pH 9.0) for anti-p53 staining. Prior to immune staining, endogenous peroxidase activity was blocked by incubating the slides in a 0.5% solution of H₂O₂ in phosphate-buffered citric acid for 15 minutes at room temperature. After washing 3 times with Tris-buffered saline (TBS) with a pH of 7.4, the sections were incubated in TBS buffer containing 10% rabbit non-immune serum (DAKO, Glostrup, Denmark) and 10% normal human plasma (and 5% bovine serum albumin [BSA] for p53 staining) for 20 minutes. Then, the sections were incubated with the primary antibody against anti-human Ki-67 antigen (clone MIB-1, DAKO) in a 1:100 dilution or p53 antigen (clone DO-7, DAKO) in a 1:100 dilution for 14 hours at 4°C. Subsequently, biotin-labeled rabbit-anti-mouse antibody (DAKO) was used as the second antibody, followed by the addition of a streptavidin-horseradish peroxidase complex (DAKO). To detect Ki-67 and p53, 3-amino-9-ethylcarbazole was used as the substrate. One experienced pathologist who was blinded to the clinical data counted at least 300 nuclei in each sample. Cells were counted as positive when moderate to intense nuclear staining was found. Staining with the isotype immunoglobulin and diluted solution was performed in the same fashion for the negative controls.

Statistical analyses were performed using Fisher’s exact test, the χ² test, Student’s t-test, the Mann–Whitney U test, and the Kruskal-Wallis test, as appropriate. A P value <0.05 was accepted as significant. Data are expressed as means ± standard deviation (SD).

POEM was successfully performed in all 15 cases without severe complications such as perforation and mediastinitis. Details of the procedure outcomes are shown in Table 2. No massive pneumoperitoneum that required reduction of abdominal distention or capnomediastinum was observed. Two patients, who had a long disease history and severe submucosal fibrosis due to preceding repeat balloon dilation, experienced post-procedural bleeding, which resolved with conservative management. The mean LES pressure decreased from 82.7 to 22.9 mmHg, and the subjective symptom scores were also significantly lower after POEM (Ekcardt score p < 0.05, dysphagia symptom score p < 0.05).

Even after a mean follow-up of 11.3 (range 8 – 25) months, no patient developed recurrent symptoms of dysphagia. Endoscopic reflux esophagitis was identified in 5 patients (33.3%) on follow-up endoscopy examination. Two (15.3%) patients received proton pump inhibitor (PPI) therapy, and the symptoms were easily controlled.

Figure 3a-d shows the endoscopic images taken from the GEJ and mid-thoracic esophagus before and after POEM. Before POEM, the endoscope was passed through the GEJ with great resistance. The thickened esophageal mucosa showed turbidity and whitish color change, lacking permeability of the vascular structures. Following POEM, however, the resistance to endoscopic passage through the GEJ was completely resolved, and the mucosal abnormalities were substantially improved, without the whitish color change. The POEM procedure allowed detailed observation of the vascular structures of the mucosal surface.

Thirteen achalasia patients underwent endocytoscopy and targeted biopsy before and after POEM. In Case 1, informed consent for the study was not obtained before the procedure, and Case 6 was excluded due to concerns about possible adverse effects of methylene blue on her pregnancy.

The average counts of esophageal mucosal nuclei on endocytoscopy before and 3 months after POEM were 128.0 (median 112.3, range 82.7-246.0) and 78.7 (median 76.0, 56.7-129.3) (Figure 4a), respectively. Thus, POEM significantly reduced the number of nuclei (p < 0.005). Compared to the endocytoscopic images obtained before POEM, cellular and nuclear structures and inter-cellular margins were observed more clearly in the images obtained after POEM (Figure 5). The nuclei numbers before POEM varied very widely, whereas they were distributed in a relatively narrow range after the procedure.

There were no significant correlation between endoscopic nuclear counts and patients’ characteristics such as degree of dilation, type, and duration of dysphagia symptom. However, statistically significant correlation between endocytoscopic nuclear counts and Eckardt score before the procedure was observed (r = 0.7937, 0.3699 ~ 0.9441, p < 0.01).

The number of nuclei in 3 non-achalasia patients (healthy controls) was also evaluated, and the average count was 88.5 (range, 76–97), which was the same level as the post-POEM values in achalasia patients.

The mean Ki-67-positive ratios before and after the procedure were 26.0 (median 25.4, 10.3-33.2) and 20.7 (median 20.0, 13.1-29.9; p = 0.07), respectively (Figure 4b). There was a significant positive correlation between the esophageal epithelial nuclei numbers assessed by endocytoscopy and the Ki-67-positive ratios (correlation coefficient, 0.695, p < 0.05) (Figure 4d).

The p53-positive ratios were decreased in the post-treatment group compared to the pre-treatment group with statistical significance. The mean positive ratios before and after the procedure were 2.35 (median 2.61, 0.32-4.23) and 0.97 (median 1.49, 0.32-1.56), respectively (Figure 4c). Histopathological evaluation with hematoxylin and eosin staining confirmed that all of the specimens were negative for dysplasia or squamous cell carcinoma.