3 per 1000 person-years, HR 1.74, 95% CI 1.29-2.35). In addition, the use of stimulants in narcolepsy patients showed lower incidence rates of bone fractures compared to non-users (incidence rates were 14.2, 11.9, and 20.0 per 1000 person-years, respectively, among frequent users, infrequent users, and non-users), but the risk estimate was not statistically significant. GDC-1971 inhibitor The evidence for associations between antidepressant use in narcolepsy patients and bone fractures was contradictory. CONCLUSION This study highlights the need to pay attention to the risk of bone fractures in narcolepsy patients, and the importance of adequate stimulants use might reduce the risk of bone fractures. OBJECTIVE Narcolepsy management usually requires lifelong pharmacotherapy. However, we know little about adherence to prescribed treatment in narcolepsy. We assessed adherence to wakefulness-promoting agents in narcolepsy patients. PATIENTS AND METHODS We retrospectively assessed adherence to wakefulness promoting medication in patients with narcolepsy using the Medicines Possession Ratio (MPR). Three levels of adherence were defined poor (≤50%), intermediate (51-79%), and good (≥80%). Refractory daytime sleepiness was defined as an Epworth sleepiness scale (ESS) score >12 despite trialling at least three wakefulness-promoting agents. We compared demographic and clinical factors, and prescribed medications between patients, stratified by levels of adherence, as well as by presence or not of refractory sleepiness. RESULTS We included 116 patients with narcolepsy (54.3% female, mean age 39.4 (±14) years). In sum, 93 (80.2%) patients had a diagnosis of narcolepsy type 1 (NT1), and 23 (19.8%) of type 2 (NT2). Suboptimal symptom control was common 39.8% had refractory sleepiness, and 47.3% of NT1 patients had persistent cataplexy. Good adherence was seen in only 55.2% of patients, while 12.9% were intermediately and 31.9% poorly adherent. Patients with poor adherence were more likely to have a diagnosis of NT2, but adherence did not vary according to gender, age, the presence of psychiatric co-morbidity, or the presence of apparent intractable symptoms. Levels of good adherence to therapy were no better in patients with refractory sleepiness than in those with satisfactory symptom control (56.5% vs 54.3%; p = 0.81). CONCLUSION Suboptimal adherence to prescribed therapy is common in narcolepsy patients, including those with apparent intractable symptoms, and particularly in patients with NT2. INTRODUCTION High-grade endometrial stromal sarcoma (HG-ESS) is a rare pathological type of uterine sarcoma. Over 80 % of affected patients would experience recurrences within a few years of initial presentation. Such case is rare and therefore, we need to report the case including the management. Information on performing good surgical techniques is important. CASE PRESENTATION A 55-year-old female patient was referred after having a total hysterectomy and bilateral salphingoophorectomy at a private hospital. Results of pathologic examination showed that the patient had HG-ESS. The patient subsequently received adjuvant chemotherapy. The regimens used were Carboplatin (AUC-6) and Paclitaxel (175 mg/m2). Within five months following the chemotherapy, she complained rapid abdominal enlargement, which was a clinically mobile palpable solid mass at the level of the navel. Imaging findings suggested recurrent endometrial stromal sarcoma in the abdominal wall. Wide excision and frozen section, which were continued with mesh insertion and abdominal wall reconstruction, were then performed. DISCUSSION Recurrence develops in one-half to two-third of patients with HG-ESS. It has been reported that the sites of recurrence usually include multiple lung metastases, peritoneal metastases, and/or local recurrences. There is currently no standard therapy for patients with recurrence of the disease as HG-ESS is a rare clinical entity. Treatment has been defined based on experiences gained from retrospective case reports. CONCLUSION Successful management of patients with recurrent abdominal wall HG-ESS requires interdisciplinary and interprofessional teamwork. Multi-centre prospective trials are required to develop a guideline of optimal treatment for the disease entity, particularly the recurrence. BACKGROUND Gastrointestinal cancer, gynecologic cancer and malignant peritoneal mesothelioma may disseminate widely to the surfaces of the abdomen and pelvis. In some patients this spread of the disease may find its way into a hernia sac. METHODS The Spigelian hernia is caused by a protrusion of the peritoneum of the anterior abdominal wall into the abdominal wall muscle. This peritoneal defect occurs through the semilunar line of Spiegel in or closely associated with the semicircular line of Douglas. RESULTS A patient presented with a large (11 cm greatest diameter) mass in the left lower quadrant. The mass was composed of malignant peritoneal mesothelioma and was anatomically located within the Spigelian hernia belt. At the time of cytoreductive surgery numerous other sites of malignant peritoneal mesothelioma were present within the abdomen and pelvis. CONCLUSIONS The left lower quadrant mass was a malignant peritoneal mesothelioma presenting within a Spigelian hernia defect. This phenomenon has not been previously reported. PURPOSE In this study we trained a deep neural network model for female pelvis organ segmentation using data from several sites without any personal data sharing. The goal was to assess its prediction power compared with the model trained in a centralized manner. METHODS Varian Learning Portal (VLP) is a distributed machine learning (ML) infrastructure enabling privacy-preserving research across hospitals from different regions or countries, within the framework of a trusted consortium. Such a framework is relevant in the case when there is a high level of trust among the participating sites, but there are legal restrictions which do not allow the actual data sharing between them. We trained an organ segmentation model for the female pelvic region using the synchronous data distributed framework provided by the VLP. RESULTS The prediction performance of the model trained using the federated framework offered by VLP was on the same level as the performance of the model trained in a centralized manner where all training data was pulled together in one centre.