A total of 149 ovarian MCTs removed from 126 females were retrieved from the surgical pathology files of the Department of Pathology, Japan Self-Defense Forces Central Hospital (Tokyo, Japan), from between November 1983 and September 2016. The clinical findings and macroscopic features were obtained from surgical pathology request forms and patient medical charts. Patients ranged in age from 14–79 years (mean, 35.1 years). Of the patients, 23 had bilateral ovarian MCTs and the other 103 had unilateral MCTs. The sizes of 144 MCTs were known, and ranged from 1–21 cm (mean, 6.6 cm). The present study was a retrospective study, which was approved by the Medical Research Ethics Committee of Japan Self-Defense Forces Central Hospital (approval number, 28–015).

All representative hematoxylin and eosin (H&E)-stained sections were histologically examined for the presence of MCT-related GI tract-like muscular walls with or without other components. In the present study, the term ‘muscular walls’ was used to describe muscularis propria-like layered aggregations of smooth muscle cells, and did not include muscularis mucosae (MM). To exclude respiratory, urogenital and uterine cervical walls, muscular walls covered by ciliated columnar, squamous and/or urothelial-like epithelium were not considered.

For all samples, 10–20% buffered formalin-fixed and paraffin-embedded representative sections containing MCT-related muscular walls were available. The fixation was performed at room temperature, ranging in duration from 12–48 h. These sections were serially cut (4–5 µm-thick), stained with H&E, and subjected to immunohistochemical analysis by incubation with antibodies against KIT (polyclonal; A4501; 1:100; Dako; Agilent Technologies, Inc., Santa Clara, CA, USA) for 60 min, cluster of differentiation (CD) 34 (NU-4A1; 413361; 1:100 dilution; Nichirei Biosciences, Inc., Tokyo, Japan) for 30 min, α-smooth muscle actin (SMA; 1A4; 412021; prediluted; Nichirei Biosciences, Inc.) for 60 min, S-100 protein (polyclonal; 422091; prediluted; Nichirei Biosciences, Inc.) for 60 min, synaptophysin (27G12; 413831; prediluted; Nichirei Biosciences, Inc.) for 30 min, discovered on gastrointestinal stromal tumors 1 (DOG1; K9; NCL-L-DOG-1; 1:100; Leica Biosystems, Newcastle, UK) for 60 min, keratin 7 (K7; OV-TL 12/30; M7018; 1:100; Dako; Agilent Technologies, Inc.) for 30 min and keratin 20 (K20; Ks 20.8; M7019; 1:100; Dako; Agilent Technologies, Inc.) for 30 min. For secondary antibody incubation, Histofine Simple Stain MAX PO (MULTI)^® (424152; Nichirei Biosciences, Inc.) was used for a duration of 30 min. Incubations with primary and secondary antibodies were performed at room temperature. Endogenous peroxidase activity and non-specific background staining was blocked with 3% hydrogen peroxidase and the blocking solution, Protein Block Serum-Free^® (X0909; Dako; Agilent Technologies, Inc.), respectively. Each blocking reagent was incubated for 30 min at room temperature before incubation with the primary and secondary antibodies.

Excluding distinct neural tissues and vessels observed on H&E-based histology, KIT⁺, CD34⁺, S-100⁺, synaptophysin⁺ and DOG1⁺ spindle or stellate cells within muscular walls were evaluated in each case, and were graded as follows: 0, no positive cells; 1, <1% positive muscular wall spindle and/or stellate cells; 2, 1–5% positive muscular wall spindle and/or stellate cells; and 3, cells >5% positive muscular wall spindle and/or stellate cells.

Clinicopathological findings were analyzed using the appropriate Spearman's rank correlation test and Fisher's exact test. Data were analyzed on a personal computer using a statistical software package (StatMate^® IV for Windows v. 4.01; ATMS Corp., Ltd., Tokyo, Japan). P<0.05 was considered to indicate a statistically significant difference.

GI tract-like muscular walls were hemi-laterally identified in 9 (7.1%) of 126 cases of MCT (4 left and 5 right MCTs). Table I summarizes the main clinicopathological findings in these 9 patients. These walls were accompanied by mucosae in 5 cases, MM-like structures in 5 cases, serosa-like features in 5 cases, and all of these components were present in 3 cases. Patients ranged in age from 19–52 years (mean, 34.3 years), and 4 patients had bilateral MCTs. The sizes of MCTs with these muscular walls ranged from 1.4–11.0 cm (mean, 4.8 cm). The largest dimension of GI tract-like structures in each case ranged from 0.2–1.5 cm (mean, 0.6 cm). GI tract-like muscular walls partially or totally demonstrated incomplete features, such as indistinct two-layered structures (all cases), thinning (7 cases) and adhesion to MM-like structures (1 case). In all muscular walls, neural tissues of varying size were observed between two conspicuous or inconspicuous muscular layers (Figs. 1 and 2), partially exhibiting Auerbach's plexus-like features, and also within the muscular layers (Fig. 2A). Gastric and intestinal mucosae were found in 2 and 4 cases, respectively. In 1 case, partially atrophic, gastric pyloric mucosa and small intestine-like mucosa were surrounded by mutual muscular walls (Fig. 1A). In another case, gastric body mucosa merged with marked flattened mucosa was observed. In 3 other cases, intestinal mucosae were difficult to distinctly divide into small and large intestinal types. Neoplastic lesions arising in GI tract-like structures, such as gastrointestinal stromal tumor (GIST), leiomyoma, glandular neoplasia and neuroendocrine tumor, were not observed.

Table II summarizes the immunohistochemical features of GI tract-like structures. In case 2, neither CD34 nor synaptophysin positivity was observed within muscular walls or the inner controls (vascular endothelium and neural cells, respectively). Hence, these findings were not included in the results. Muscular walls and MM-like structures were highlighted by SMA positivity, but the SMA positivity of the former was relatively weak in 3 cases (Fig. 1B). A few weakly KIT⁺ spindle/stellate cells (not including KIT⁺ mast cells) were observed chiefly nearby neural tissues between muscular layers in 7 cases (Figs. 1D and 2B), and were graded 1 and 2 in 5 cases and 2 cases, respectively. S-100⁺ spindle or stellate cells were scattered within muscular layers and within the distinct neural tissues recognizable on H&E-stained histology in all cases (Figs. 1E and 2C). These intramuscular S-100⁺ cells were considered to be richly distributed neural cells or networks, and were graded 1, 2 and 3 in 3 cases each. CD34⁺ spindle or stellate cells were scattered near the boundary zones of two muscular layers and within muscular layers in 8 cases (Figs. 1F and 2D) along with or without prominent CD34⁺ microvessels, and were graded 1, 2 and 3 in 3, 1 and 4 cases, respectively. Other than in the context of distinct neural tissue positivity, intramuscular synaptophysin⁺ spindle or stellate cells were observed in 7 cases (Fig. 2E) and were graded 1, 2 and 3 in 3, 2 and 2 cases, respectively. The intramuscular distributions of KIT⁺, CD34⁺, S-100⁺ and synaptophysin⁺ spindle or stellate cells were similar. A few DOG1⁺ spindle or stellate cells were observed near the Auerbach's plexus-like neural tissues in only 1 case (case 6; Fig. 2F), possibly corresponding with KIT⁺ cells. Accompanying gastric mucosae contained only K7⁺/K20⁻ cells in 1 case, and both K7⁺/K20⁻ and K7⁻/K20⁺ cells in 1 case. Intestinal mucosa specimens were composed of only K7⁻/K20⁺ cells in 2 cases, and were admixed with K7⁺/K20⁻ cells and K7⁻/K20⁺ cells in 2 cases. Serosa-like structures were positive for K7 (Fig. 1C).

The presence of GI tract-like muscular walls was not statistically associated with age (P=0.855), MCT size (P=0.082) or bilateral presence of MCTs (P=0.096; Table III). Spearman's rank correlation analysis demonstrated significant correlations between grading scores of KIT⁺ and S-100⁺ cells (P=0.014), CD34⁺ and S-100⁺ cells (P=0.003), CD34⁺ and synaptophysin⁺ cells (P=0.009) and S-100⁺ and synaptophysin⁺ cells (P=0.002); however, no significant correlations were observed between other combinations (Table IV). Fisher's exact test (data not shown) revealed significant relationships between CD34⁺ and S-100⁺ cells (P=0.018) and between S-100⁺ and synaptophysin⁺ cells (P=0.048); however, it revealed no significant associations between KIT⁺ and S-100⁺ cells (P=0.500) or between CD34⁺ and synaptophysin⁺ cells (P=0.143). No statistical associations were observed between decreased SMA positivity of muscular walls and other immunostaining results.