Dynamics of thoracic endometriosis in the pleural cavity

Catamenial pneumothorax (CP) is defined as recurrent secondary spontaneous pneumothorax, occurring during menstruation in reproductive-aged women. It was reported that the incidence of endometriosis-related pneumothorax occurring in the intermenstrual period, ranges from 38% to 63%, owing to endometriotic lesions in both the visceral and parietal pleura as well as the diaphragm [ 1 , 2 ]. By its strict definition, CP does not include pneumothorax caused by thoracic endometriosis in the intermenstrual periods and includes primary spontaneous pneumothorax occurring in the menstrual period. Therefore, some studies had used the term thoracic endometriosis-related pneumothorax (TERP) to describe the essential pathophysiology of this disease [ 2 – 4 ]. In general, endometriosis affecting the diaphragm is thought to be related to the development of TERP [ 5 – 9 ]. However, several studies have reported that TERP patients could also present with visceral and/or parietal pleural endometriotic lesions [ 1 , 10 – 12 ]. Diaphragmatic lesions could cause pneumothorax through the diaphragm passage [ 5 , 11 , 13 ] and visceral pleural lesions could cause pneumothorax through the pleural passage [ 6 , 14 ]. Parietal pleura lesions, on the other hand, may not be directly related to pneumothorax, but possibly disseminate in the thoracic cavity, such as the visceral pleura. Although surgery was regarded as effective treatment for TERP, postoperative recurrence rates have remained high [ 1 , 9 – 11 , 15 – 18 ]. This is probably because we have not adequately understood the essential pathophysiology of thoracic endometriosis, including diaphragmatic, visceral pleural, and parietal pleural lesions. Therefore, we aimed to clarify the clinicopathological features of thoracic endometriosis and the dynamics of the endometriotic tissues in the pleural cavity from our cases of TERP.

The patient characteristics and clinical background of the enrolled 160 patients were described in Table 1 . All patients were non-menopausal females. Most cases had right-sided pneumothorax with only one case having a left-sided pneumothorax. Fourteen patients were histologically and surgically diagnosed with pelvic endometriosis and 71 were suspected to have pelvic endometriosis by transvaginal sonography and/or magnetic resonance imaging. SD, standard deviation. All the 160 patients had diaphragmatic endometriotic lesions. Endometriosis in the visceral pleura was diagnosed in 79 patients (49.4%), and endometriosis in the parietal pleura in 71 patients (44.4%). Thirty-six patients had both the visceral and parietal endometriotic lesions. Multiple lesions were confirmed in some patients in both the visceral and parietal pleura. Regarding visceral pleural lesions, one lesion was confirmed in 61 patients, two lesions were in 11, and three lesions were in 7. These added up to 104 lesions in total. Regarding parietal pleural lesions, one lesion was confirmed in 51 patients, two lesions were in 12, three lesions were in 6, and four lesions were in 2. These added up to 101 lesions in total. Visceral pleural endometriosis showed more cystic lesions rather than blueberry spots. The number of positive cases for cystic lesions, blueberry spots, and combined cases were 80 (76.9%), 14 (13.5%), and 10 (9.6%), respectively. The distribution of visceral and parietal pleural lesions under thoracoscopy was shown in Fig 2 . For visceral pleural lesions, 66 lesions (63.5%) were detected in S 4 , 13 lesions (12.5%) in S 6 , and 7 lesions (6.7%) in S 2 , respectively. Eighty-six lesions (82.7%) were found concentrated in the area where the upper, middle, and lower lobes of the lung intersect. Among these, eighty-two lesions (82/86: 95.3%) existed within the interlobar surface or at the border of each segment. In the parietal pleural lesions, 25 lesions (24.8%) were detected in the dorsal 6 th intercostal space (ICS). Eighty-seven lesions (86.1%) were found in the dorsal 4 th -9 th ICS. (A) Visceral pleural lesions, and (B) Parietal pleural lesions. Eighty-six visceral pleural lesions (82.7%) out of 104 were observed in S 4 , S 6 , and S 2 . Eighty-seven parietal pleural lesions (86.1%) out of 101 were observed in 4 th -9 th dorsal intercostal space. Endometriotic stromal cells were confirmed in all 160 (160/160: 100%) diaphragmatic lesions, and endometriotic gland cells were found in 61 (61/160: 38%). In 104 visceral pleural lesions, endometriotic stromal cells were confirmed in all lesions (104/104: 100%). In contrast, endometriotic gland cells were confirmed in only three (3/104: 2.9%). In 101 parietal pleural lesions, endometriotic stromal cells were confirmed in 100 lesions (100/101: 99%) and endometriotic gland cells were confirmed in 17 lesions (17/101: 17%). All endometriotic stromal tissues were stained positive for ER, PgR, and CD10, and all endometriotic gland tissues were stained positive for ER and PgR, but negative for CD10. We then histopathologically investigated in greater detail, the 104 visceral pleural lesions, paying close attention to location and tissue invasive behavior. Some endometriotic tissues stayed on visceral pleural surfaces, while others demonstrated inward tissue invasion from the visceral pleural surfaces ( Fig 3 ). The depth of invasion varied per lesion, with some reaching up to the subpleural layer or pulmonary alveolus; however, all lesions appeared to exist continuously from the outer side of the external elastic layer of the visceral pleura. Endometriotic tissues localized in the visceral pleural surface were confirmed in 54 lesions (51.9%), tissues invading the external elastic layer in 20 (19.2%), and tissues invading the internal elastic layer in 30 (28.8%) ( Table 2 ). (A) Endometriotic tissue adhered to the surface of the visceral pleura. Note that marked hemosiderin deposition was observed in the entire pleural wall. (Hematoxylin-Eosin stain, inset: the external and internal elastic layer were indicated by black arrow and arrowhead, respectively (Elastica von Gieson stain)). (B) Endometriotic lesions surrounded by granulation and/or fibrous tissue accompanied with inflammatory cells invaded into the lung tissue through the visceral pleura. (Hematoxylin-Eosin stain, inset: white arrow and arrowhead indicated the break of the external and internal elastic layer caused by the invasion of endometriotic tissue, respectively (Elastica von Gieson stain)). (C) Endometriotic stromal cells in the endometriotic lesion were immune-positive for CD10 (gray arrows) in the same tissue of Fig 3B . Note that the anti-CD10 antibody was cross-reactive to alveolar stromal cells (black arrowhead) and alveolar macrophages (white arrowhead). The median follow-up period was 370 (range 6–1824) days. A total of 30 patients (30/160: 18.8%) developed postoperative recurrences. The mean number of recurrences was 3.1±3.7. The median time to recurrence after surgery was 195 (range 23–1303) days. Twenty-two patients (22/30: 73.3%) developed recurrences within 2 years. The Kaplan-Meier method revealed that the 1- and 2-year postoperative recurrence rates were 13.8% and 19.3%, respectively ( Fig 4 ). Of the 30 patients with postoperative recurrences, 24 received no additional treatments and were followed up: 10 had one recurrence and have not relapsed since, whilst 14 had multiple recurrences; however, postoperatively the pneumothorax developments were noted to be reduced in both frequency and intensity ( Fig 5 ). The other 6 patients underwent reoperations because the rate of pneumothorax developments did not decrease post-surgery. The frequency of recurrences was determined by dividing the total number of onsets for all patients by their total observation period, according to the previous report [ 19 ]. Using this calculation method, the number of cumulative recurrences were 40 and 54, and the frequencies of cumulative recurrences were 4.51 and 1.53 times per year in the patients who underwent reoperation and those who did not, respectively. In those who underwent reoperation, all visceral pleural endometriotic lesions were confirmed in S 4 , S 6 , and S 2 (5/5: 100%) and majority of the parietal pleural lesions were in the dorsal 5 th -7 th ICS (3/5: 60%). In those who did not undergo reoperation, most of the visceral pleural lesions were confirmed in S 4 , and S 6 (12/13: 92%) and majority of the parietal pleural lesions were in the dorsal 4 th -8 th ICS (13/18: 72%). In addition, the detailed characteristics of the 6 patients who received reoperations were shown in Table 3 . We resected any suspicious lesions and performed pleural covering of the specific sites of thoracic endometriosis with ORC for all cases. The residual endometriosis was histopathologically confirmed in 3 cases (3/6: 50%); the locations of endometriosis were diaphragm, S 1 , S 4 in the visceral pleura, and dorsal 4 th ICS in the parietal pleura. Of the 6 patients who underwent reoperations, 4 patients have not relapsed. Of the 2 cases of relapse, one had 2 recurrences following reoperation and has not relapsed since, whilst the other had 1 recurrence, has started hormone therapy (Dienogest) and has not relapsed since. † This patient had two recurrences and has not relapsed since. ‡ This patient had one recurrence; following the administration of Dienogest, has not relapsed since. D: diaphragm; V: visceral pleura; P: Parietal pleura; ICS: intercostal space.

The ethical committee at Nissan Tamagawa Hospital approved this retrospective study (No. 2019–032), and the need for informed consent from each individual was waived with a choice to opt out. In our hospital, 435 female patients aged 55 and younger underwent video assisted thoracoscopic surgery under one-lung ventilation in the lateral decubitus position for various types of pneumothoraxes from January 2015 to December 2019. A total of 166 patients, who were suspected of having thoracic endometriosis after pleural cavity inspection, underwent video assisted thoracoscopic surgery for resection of thoracic endometriosis. Among these, 160 patients were histopathologically diagnosed with thoracic endometriosis post-procedure. The following clinicopathological features were assessed by retrospective chart reviews: age at pulmonary surgery, laterality of pneumothorax, history of pelvic endometriosis, smoking history, and body mass index. History of pelvic endometriosis was confirmed by interview, transvaginal sonography, and pelvic magnetic resonance imaging. In the surgical procedure, the visceral and parietal pleura, in addition to the diaphragm, were inspected carefully to identify signs of thoracic endometriosis. The diaphragm was partially resected by the stapling method and repaired by the hand suture method when diaphragmatic endometriosis was suspected. Cystic lesions in the visceral pleura, as well as the nodule-like hematomas called blueberry spots, were removed by stapling or after ligation ( Fig 1A and 1B ). Red or brown nodules and dents of the pleura called depressed lesions in the parietal pleura, including the intercostal muscle, were removed by partial pleurectomy ( Fig 1C and 1D ). All lesions were resected with careful attention to the resection margin. The areas which were partially resected owing to suspicion of thoracic endometriosis were reinforced with oxidized regenerated cellulose (ORC) in order to prevent the recurrence of pneumothorax and pleural adhesion. Visceral pleural endometriosis showed (A) brown nodule, and (B) cystic lesion. Brown nodule was like a hematoma, to say it in another way blueberry spot. Parietal pleural endometriosis showed (C) brown nodule, and (D) dent of the pleura called depressed lesion. All surgical specimens were assessed by hematoxylin-eosin staining. Moreover, immuno-histological examinations were performed using antibodies against estrogen receptor (ER), progesterone receptor (PgR), and CD10. All endometriotic lesions were diagnosed by confirming the presence of the endometriotic stromal or gland cells in the surgical specimens via pathological examination. Particularly with regards to endometriotic stromal cells, positive immunostaining of ER and PgR as well as CD10 could lead to a diagnosis. In addition, we focused on visceral pleural lesions to research the mechanism of pneumothorax in visceral pleura-associated endometriosis in detail. The structure of the visceral pleura is composed of five layers: mesothelial cells, sub-mesothelial, external elastic, subpleural, and internal elastic. We microscopically evaluated the invasion level of endometriotic tissues in the visceral pleura using Elastica Van Gieson staining. All histopathological evaluations were performed by a pathologist. Recurrence was defined by collapse of the lung, after confirming the complete expansion of the ipsilateral side postoperatively by chest radiography or computed tomography. Postoperative symptoms relating to thoracic endometriosis, such as pain, were not included in this definition as the most common reason that the patients enrolled in this study had received surgery was not such symptoms but pneumothorax. Reoperation with positive findings of thoracic endometriosis was also not included, as patients with pneumothorax recurrence did not always undergo reoperation. The Kaplan-Meier method was used to assess the postoperative recurrence rate, utilizing the statistical software program “R” Ver 3.3.2 (R Foundation for Statistical Computing, Vienna, Austria).

The visceral pleural lesions in TERP could mainly develop from pleural dissemination and infiltration into the lung. It may be essential to make careful observation of the following specific sites for thoracic endometriosis: visceral pleura in S 4 , S 6 , S 2 , and parietal pleura in dorsal ICS, because visceral and parietal pleural endometriosis, in addition to diaphragm endometriosis, could participate in postoperative recurrence of TERP.

(XLSX) Click here for additional data file.