Dynamics of diaphragmatic endometriosis with thoracic endometriosis-related pneumothorax

Endometriosis is defined as the presence of endometrial tissue outside of the uterus and affects 6% to 10% of women of reproductive age ( 1 ). Thoracic endometriosis is a disease with pathophysiological conditions that are different from those of endometriosis and is considered a type of extra-pelvic endometriosis ( 2 ). Pneumothorax caused by thoracic endometriosis is referred to as thoracic endometriosis-related pneumothorax (TERP) ( 3 , 4 ). TERP has a high postoperative recurrence rate; therefore, it often remains unresolved. Because air in the thoracic cavity is thought to enter through the diaphragm via the abdominal cavity or lungs, our research center elucidated the dynamics of thoracic endometriosis in the lungs and chest wall ( 5 , 6 ). Therefore, this study aimed to clarify the dynamics and mechanisms of metastasis to the diaphragm associated with TERP. We present this article in accordance with the STROBE reporting checklist (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-838/rc ).

The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of Nissan Tamagawa Hospital (No. 2023-023), and an opt-out option for patients was offered. We obtained informed consent from each patient by opt-out. Of the 140 patients who underwent initial video-assisted thoracic surgery (VATS) for TERP at our hospital between January 2018 and December 2023, 130 were enrolled in this study. Ten patients were excluded (six patients with adhesions, one patient with a diaphragmatic hernia, two patients without diaphragmatic lesions, and one patient with missing surgical image data) ( Figure 1 ). Age, smoking history, preoperative hormone therapy, and diagnostic data regarding pelvic endometriosis were collected from the medical records ( Table 1 ). Diaphragmatic lesions were morphologically classified as hiatus, hematoma, and scar types ( Figure 2 ). The diaphragmatic area was virtually divided into four equal parts by drawing lines parallel and perpendicular to the hepatic coronary ligament at the tendon center ventral to the hepatic coronary ligament. The ventral lateral area was considered area A, and the other areas (proceeding in the counterclockwise direction) were considered areas B, C, and D. The area dorsal to the right triangular ligament and coronary ligament was considered area E ( Figure 3 ). Lesion distribution was evaluated based on data from the surgical reports and images obtained during surgery. Flowchart of the study participants. Data are presented as median [range] or n [%]. CA125, cancer antigen 125. Diaphragmatic lesions (arrows): (A) hiatal type; (B) hematoma type; and (C) scar type. Diaphragm areas. Patients were placed in the left lateral position and three-port VATS was performed under general anesthesia with separated lung ventilation. One port was inserted at the fifth intercostal space on the mid-axillary line. Two ports were inserted 3 cm apart at the intersection of the mid-axillary line and the level of the diaphragm, lateral to the planned resection line. A 10-mm flexible thoracoscope (Olympus, Tokyo, Japan) was used for the procedure. The diaphragmatic lesion was resected using Endo GIA Ultra Universal Staplers (Medtronic, Dublin, Ireland) to achieve hemostasis. The defect was closed using the hand suture method with 2-0 Vicryl (Ethicon, Cincinnati, OH, USA). After the lung lesions were ligated with the PDS-ENDOLOOP (Ethicon), partial resection was performed. All diaphragmatic, pulmonary, and chest wall lesions were stained with hematoxylin and eosin and anti-estrogen, anti-progesterone, and anti-CD10 antibodies after resection. Statistical analyses were performed using the EZR software package ( http://www.jichi.ac.jp/saitama-sct/SaitamaHP.files/statmed.html ). Continuous variables (presented as medians and interquartile ranges) were evaluated using the Mann-Whitney U test. Categorical variables were evaluated using Fisher’s exact test. For each test, P<0.05 was considered significant.

Endometrial stromal tissue was observed in all diaphragmatic lesions ( Table 1 ). Eleven patients (8%) had undergone surgery for pelvic endometriosis prior to TERP, with the diagnosis confirmed by pathological examination. 13 patients (29%) were clinically diagnosed with pelvic endometriosis based on gynecological evaluation, including pelvic MRI, transvaginal ultrasound, or bimanual examination. Fifteen patients (39%) had received hormone therapy prior to surgery. Endometrial lesions were observed in area A in 130 patients (100%), area B in 52 patients (40%), and area D in 21 patients (16%); additionally, three lesions were observed in area C (2%). Hematoma and scar lesions were found in area E (2%) ( Table 2 ). Lesions in areas A, B, and D were located along the border between the membrane and muscle of the diaphragm (95.8%) ( Table 3 ). Hiatus, hematoma, and scar lesions were observed in areas A, B, and D; however, the most (60%) were observed in area A. No hiatus lesions were found in area E ( Table 4 ). In areas A, B, and D, 99% of the lesions were hiatus-type; these were observed along the border between the membrane and muscle of the diaphragm ( Table 4 ). Scar lesions (two cases) and hematoma lesions (one case) were observed in area E; the pathological diagnosis revealed that these lesions were only present on the intrathoracic side of the diaphragm and were not on the retroperitoneal side of the diaphragm ( Figure 4 ). During this study, 180, 60, and 56 hiatal, hematoma, and scar lesions, respectively, were observed. Microscopic findings of lesions in area E (magnification: ×400). (A) Endometrial stroma is detected in the resected specimens (hematoxylin-eosin staining). (B) Endometrial stroma exhibits nuclear staining for estrogen receptors (immunohistochemistry). (C) Endometrial stroma exhibits nuclear staining for progesterone receptors (immunohistochemistry). (D) Endometrial stroma exhibits nuclear staining for CD10 (immunohistochemistry).

Three theories, the transplant theory, lymphohematogenous metastasis theory, and metaplasia theory, regarding the mechanism of diaphragmatic endometriosis exist. The transplant theory, which was proposed by Sampson et al. , is considered the most likely explanation of endometrial tissue flow into the thoracic cavity ( 7 ). According to this theory, endometrial tissue invades the pelvis through the reflux of menstrual blood and attaches to the right diaphragm by following the physiological clockwise flow of ascites. Maulitz et al. reported that endometriosis that develops in the bronchi is caused by lymphohematogenous metastasis ( 8 ). The metaplasia theory hypothesizes that endometriosis develops when mesothelial cells that cover the surface of the pleura and peritoneum transform into endometrial glandular or stromal tissue ( 8 ). Nezhat et al. explained this hypothesis by showing that endometriosis develops in patients with Mayer-Rokitansky syndrome who lack a functional endometrium and in men who are administered high-dose estrogen ( 9 ). Left diaphragmatic lesions were not observed in the 130 patients with TERP in this study; all lesions were right diaphragmatic lesions. Therefore, the metaplasia theory and lymphohematogenous metastasis theory were considered unacceptable for these patients. Additionally, the right diaphragmatic lesions could not be explained by the transplant theory, which hypothesizes that the flow of ascites is the only contributing factor. Vercellini et al. reported that endometrial tissue in the diaphragm was observed in the abdominal cavity during laparotomy or laparoscopic surgery for pelvic endometriosis; additionally, this tissue was observed on the right side in 66% of patients, on the left side in 6% of patients, and on both sides in 27% of patients ( 10 ). Fukuoka et al. , Ochi et al. , and Ceccaroni et al. each also reported that endometrial tissue can easily attach to the right diaphragm ( 3 , 5 , 11 ). The pocket theory and shelter effect can explain why thoracic endometriosis with right-side pneumothorax and endometrial tissue attachment to the right diaphragm are common. The area surrounded by the peritoneal surface of the diaphragm, surface of the right lobe of the liver, right triangular ligament, hepatic coronary ligament, and hepatic falciform ligament forms a pocket; additionally, the endometrial tissue floating in the abdominal cavity enters the pocket and stagnates at the dead end of the anatomy. We named it “pocket theory”. In the space between the diaphragm and liver, a small triangular gap comprising the center of the diaphragm tendon, liver surface, and thick diaphragm muscle exists ( Figure 5 ). Berlanda et al. reported that endometrial tissue becomes established and grows in a stable environment ( 12 ). This triangular gap is a stable shelter. A continuous triangular gap (areas A, B, and D) exists at the diaphragm tendon along the border between the tendon and muscle of the diaphragm. We believe that the concentration of diaphragmatic endometrial tissue is caused by the presence of this gap. However, the shelter effect does not occur at area C because a gap does not exist at that site. Gross findings of the diaphragmatic peritoneum. During thoracoscopic surgery, we incise the diaphragm and observe the peritoneal surface of the diaphragm; resection is performed if endometrial tissue is present ( Figure 6 ). Because the diaphragm muscle is thick, endometrial tissue cannot penetrate the thoracic cavity; however, because the diaphragm membrane is thin, the endometrial tissue can easily invade the thoracic cavity. Gross findings of lesions in area E. Area A is the most frequent site of diaphragmatic endometriosis because the ventral lateral side of the diaphragm is motionless and stable. Hao et al. reported that the ventral lateral side of the diaphragm has little respiratory movement and that the shelter effect is most visible in area A ( 13 ). The endometrial tissue in area E was on the side of the thoracic cavity. Area E is the retroperitoneal site and is not anatomically invaded from the site of retroperitoneal cavity ( Figures 4,7 ). Because there is no endometrial tissue in the retroperitoneal space. Endometrial tissue invasion pathways. Ceccaroni et al. reported that all of their patients (n=215) had lesions on the thoracic side, and that 133 (62%), 58 (27%), and 24 (11%) had hiatus, hematoma, and scar lesions, respectively ( 11 ). Hiatus lesions form during a process whereby the endometrial tissue becomes established and proliferates in the diaphragm ( 14 ). The proportions of the hiatus, hematoma, and scar lesions observed during the aforementioned study were similar to those observed during our study ( Table 4 ). Although it cannot be completely ruled out that the intrathoracic scar lesions observed in this study may have been caused by factors other than TERP, based on clinical findings and previous reports, we considered them to be associated with TERP. While our clinical experience suggests that patients with scar formation are likely to have received hormone therapy prior to surgery, the data from this study did not demonstrate a clear association between hormone therapy and scar formation. Hiatus lesions are unique to the diaphragm and are not found with other types of endometriosis. Endometrial stromal tissue plays an important role in endometriosis invasion, which is promoted by epithelial-mesenchymal transition (EMT) in endometrial tissue ( 15 ). Matrix metalloproteinases (MMPs) and other proteolytic enzymes are involved in EMT. The role of MMPs in decomposing the extracellular matrix has attracted attention because they are involved in metastasis of cancer tissue and invasion of target organs ( 16 ). Furthermore, endometrial tissue is involved in metastasis, engraftment, invasion, and proliferation because EMT is induced by MMPs. When endometrial tissue engrafts and proliferates at the membrane of the diaphragm tendon, MMPs dissolve and weaken the surrounding structures. The diaphragm membrane comprises connective tissue fibers with a lattice shape, and the diaphragm muscle repeatedly contracts and relaxes during long-term breathing movements. The diaphragm membrane is subjected to strong tension in the radial and circumferential directions during contraction. Therefore, contraction and relaxation of the diaphragm cause cracks and gaps among the radial fibers, and the endometrial tissue likely invades the cracks to form a hiatus lesion. The annual number of surgical TERP cases in Japan is approximately 200 because of its rarity ( 17 ). Therefore, only case reports are available. Currently, prospective studies of rare diseases are limited by the number of cases encountered at each facility, and detailed observations and investigations are insufficient, thus creating challenges in obtaining research results. Additional knowledge may be obtained by conducting retrospective studies at a single facility with a large number of cases. Our research center has encountered more than 600 cases of TERP within the past 15 years, thus allowing a detailed examination of its pathology, which was the main focus of this study. Studies including a high-volume center with a large number of cases are important to elucidating the pathology and dynamics of TERP. However, studies of TERP are very limited because it straddles the fields of gynecology and thoracic surgery and its rarity. Additionally, information-sharing has been limited because few researchers have focused on TERP. Therefore, we believe that great progress in elucidating the pathophysiology and treatment of TERP is possible if gynecologists, thoracic surgeons, and pathologists collaborate and conduct independent research of endometriosis with different pathologies at various sites.

As explained by the pocket theory and shelter effect, diaphragmatic lesions associated with TERP are most commonly observed on the ventral side of the diaphragm membrane. The synergistic effects of physical traction caused by respiratory movement of the diaphragm and the protein dissolution phenomenon observed with endometrial tissue invasion are important to the formation of hiatus lesions in the diaphragm membrane.

The article’s supplementary files as