All That Is Small Is Not a Small-Cell Carcinoma: Thoracic SMARCA4-Deficient Undifferentiated Tumors Masquerading as SCLC.

Small-cell lung carcinoma (SCLC) cell lines have been widely utilized as a preclinical model of this highly aggressive disease. However, since their creation decades ago, novel tumor entities have been defined that might clinicopathologically mimic SCLC, which notably includes thoracic SMARCA4-deficient undifferentiated tumor (SMARCA4-UT). Multiomic reassessment of the presumed SCLC cell lines with high YAP1 expression reveals that nearly all of these tumors represent unsuspected SMARCA4-UT. See related article by Ng et al., p. 1846.

Thoracic SMARCA4-deficient undifferentiated tumor: current knowledge and future perspectives.

Thoracic SMARCA4-deficient undifferentiated tumor is a newly recognized disease entity characterized as a high-grade malignant neoplasm with an undifferentiated or rhabdoid phenotype. The tumor was initially identified as a subtype of thoracic sarcoma with SMARCA4 loss, but further investigation resulted in its classification as a subtype of epithelial malignancies in the current World Health Organization classification. SMARCA4-deficient undifferentiated tumor is highly aggressive and has a poor prognosis. Because of its rarity, an optimal treatment strategy has not yet been identified. In this review, we summarize the literature on SMARCA4-deficient undifferentiated tumor in terms of its clinical characteristics, diagnosis, treatment strategy and future perspectives.

NUT carcinoma and thoracic SMARCA4-deficient undifferentiated tumour: facts and controversies.

NUT carcinoma and thoracic SMARCA4-deficient undifferentiated tumour are unique entities in the 5th edition of the World Health Organisation (WHO) Classification of Thoracic Tumours, whose definitions include molecular genetic abnormalities. These aggressive tumours require rapid work-ups on biopsies, but a broad list of differential diagnoses poses challenges for practising pathologists. This review provides an update on their key clinicopathological and molecular characteristics, as well as controversies regarding tumour classification and diagnostic strategy. Phenotypical assessment plays a substantial role in diagnosis because recurrent and predictable clinicopathological findings exist, including robust immunohistochemical phenotypes. Accurate diagnosis is crucial for appropriate management and a clearer understanding of the disease.

[Chest wall repair after extensive resections]. / Rezul'taty vosstanovleniya karkasa grudnoi stenki posle obshirnykh rezektsii.

OBJECTIVE: To analyze the outcomes after different methods of post-resection chest wall defect reconstruction. MATERIAL AND METHODS: The study included 41 patients aged 22-73 years who underwent chest wall repair with local tissues and synthetic materials. Twelve (29.3±7.1%) patients had sarcoma, 9 (21.9±5.9%) - non-small cell lung cancer (NSCLC) with invasion of the chest, 9 (21.9±5.9%) - metastatic lesions, 8 (19.5±6.2%) - benign tumors, 2 (4.8±3.4%) - breast cancer with invasion of the chest wall, 1 (2.4±2.4%) - desmoid tumor. Seven patients were diagnosed with T3N0M0, 1 - T3N2M0, 1 - T2N0M1b (oss). Among patients with NSCLC with invasion into the chest wall, squamous cell cancer was verified in 4 (44.4±16.6%) patients, adenocarcinoma - in 4 (44.4±16.6%), neuroendocrine tumor - in 1 (11.2±10.5%) patient. Stages of surgeries are presented. RESULTS: We analyzed treatment outcomes in 41 patients. Five (12.2%) patients had seroma, hemothorax, thoracopleural fistula, subcutaneous emphysema and fatal asystole. There were no postoperative complications associated with paradoxical breathing. CONCLUSION: Accurate morphological verification prior to treatment is valuable to determine the stages of combined treatment of chest wall tumors. Chest wall defect closure with own tissues and synthetic materials is necessary after extensive resections. A multidisciplinary approach involving thoracic and plastic surgeons is needed.

Suitability of respiratory endoscopy for sampling malignant thoracic tumors for comprehensive genomic profiling.

Comprehensive genomic profiling (CGP) is implemented to detect actionable gene aberrations and design matched therapies. Although malignant thoracic tumors are commonly detected through respiratory endoscopy, it is questionable whether the small specimens obtained thereof are sufficient for CGP. Therefore, this study aimed to investigate the suitability of respiratory endoscopy for sampling primary and metastatic thoracic tumors for CGP. Patients whose specimens were collected through respiratory endoscopy and assessed by pathologists to determine their suitability for CGP at our institution between June 2019 and May 2022 were reviewed retrospectively. The suitability of each procedure as a sampling technique for CGP and, in the cases actually analyzed, the distribution of the detected gene aberration were assessed. In total, 122 patients were eligible for analysis; the median age was 61 (range, 29-86) years, and 71 (58.2%) patients were male. Primary intrathoracic tumors were found in 91 (74.6%) cases, including 84 (68.9%) primary lung cancers; the remaining thoracic metastases of extrathoracic origin included various types. The suitability rates of specimens obtained using conventional bronchoscopy with and without cryobiopsy, endobronchial ultrasound-guided transbronchial needle aspiration, and medical thoracoscopy were 82.8% (24/29), 70.4% (19/27), 72.9% (35/48), and 100% (18/18), respectively. Of the 96 cases judged suitable, 83 were subjected to CGP, and all but one were successfully analyzed. Finally, 47 (56.6%) patients had at least one actionable gene aberration and eight (9.6%) were treated with the corresponding targeted therapies. In conclusion, specimens obtained through respiratory endoscopy are suitable for CGP; medical thoracoscopy and cryobiopsy in conventional bronchoscopy are particularly useful.

[Feasibility of cardiac surgical techniques for intrathoracic tumors]. / O tselesoobraznosti ispol'zovaniya kardiokhirurgicheskikh tekhnologii pri vnutrigrudnykh opukholyakh.

OBJECTIVE: To analyse safety and expediency of cardiac surgical technologies including cardiopulmonary bypass (CPB) in patients with locally advanced lung cancer and invasive tumors of the mediastinum. MATERIAL AND METHODS: Cardiac surgical techniques and CPB were used in 23 patients (group 1) with locally advanced thoracic tumors between 2005 and 2015. For the same period, there were 22 patients (group 2) who underwent combined surgeries and could have had similar techniques. However, these techniques were not used for various reasons. Mediastinal malignancies and non-small cell lung cancer were diagnosed in 26 (57.8%) and 19 (42.2%) patients, respectively. Invasion of superior vena cava (n=15), aorta (n=13) and pulmonary artery (n=12) was the most common. Lesion of innominate vein (n=8), left atrium (n=6) and innominate artery (n=4) was less common. A total of 21 pneumonectomies were performed (14 in the first group and 7 in the second group). Lobectomy was less common (one patient in each group). Sublobar lung resection was performed in 10 patients (2 patients in the first group and 8 ones in the second group). All resections were total in the first group (R0) that was confirmed by routine morphological examination of resection margins of different organs and vessels. The situation was worse in the second group (R1 in 19 (86.4%) patients, R2 in 3 (13.6%) patients). RESULTS: Total postoperative morbidity was 53.3%, mortality - 8.2%. These values are higher compared to patients undergoing surgical treatment for thoracic malignancies. Incidence of postoperative complications was higher in the first group (16 (69.6%) and 8 (36.4%), respectively). Four patients died in the first group. Sepsis (n=2), acute right ventricular failure (n=1) and acute myocardial infarction (n=1) caused death. There were no lethal outcomes in the second group. Various postoperative complications were diagnosed only in 8 (36.4%) patients. The long-term results were followed-up in 80% of patients. In the first group, 3- and 5-year survival rates were 30.5% and 25%, respectively (median 43.8 months). In the second group, these values were 25% and 2%, respectively (median 24.9 months). Long-term mortality in the second group was caused by progression of malignant process, including local recurrence, after palliative surgery (R1, R2 resection). CONCLUSION: Higher risk of postoperative complications and mortality in patients undergoing on-pump surgery is compensated by significantly better long-term results. Further progress is associated with higher safety of CPB, as well as solving some organizational and educational problems.

Diagnosis of thoracic SMARCA4-deficient undifferentiated tumor in cytology.

INTRODUCTION: Although alterations in SMARCA4-deficient occur in non-small cell lung carcinoma (SD-NSCLC), thoracic SMARCA4-deficient undifferentiated tumor (TSDUT) is recognized as a distinct entity in the 2021 World Health Organization Classification of Thoracic Tumors because of unique morphologic, immunophenotypic and molecular features, and worse survival compared with SD-NSCLC. Cytologic diagnosis of TSDUT is clinically important because of its aggressive behavior and because it is often diagnosed by fine-needle aspiration because TSDUTs are usually unresectable at presentation. Here, we identify cytologic features that can be used for recognition of TSDUT and distinction from SD-NSCLC. MATERIALS AND METHODS: Cytomorphologic features were investigated in cytology specimens from patients with TSDUT (n = 11) and compared with a control group of patients with SD-NSCLC (n = 20). RESULTS: The presence of classic rhabdoid morphology, at least focally, was entirely specific for TSDUT (n = 6, 55%) compared with SD-NSCLC (n = 0) in this study. TSDUT more frequently showed tumor necrosis (n = 11, 100% vs. n = 8, 40%; p = .001), dominant single-cell pattern on aspirate smears or touch preparation slides (n = 8 [of 9], 80% vs. n = 3, 15%; p = .010), nuclear molding (n = 5, 45% vs. n = 1, 5%; p = .013), and indistinct cell borders (n = 11, 100% vs. n = 5, 25%; P < .001) compared with SD-NSCLC, respectively. CONCLUSIONS: Cytomorphologic features occurring more frequently in TSDUT include tumor necrosis, dominant single-cell pattern, nuclear molding indistinct cell borders, and focal rhabdoid cells. Presence of these features in a cytology specimen of an undifferentiated tumor, particularly in a patient with a thoracic mass, should raise suspicion for TSDUT and prompt appropriate ancillary workup.

[Thoracic SMARCA4-deficient undifferentiated tumor-pathological diagnosis and combined immune checkpoint inhibitor treatment].

We explored clinicopathological features and treatment strategies for thoracic SMARCA4-deficient undifferentiated tumor (SMARCA4-UT). Thoracic SMARCA4-UT is a new entity recently acknowledged in the 2021 edition of World Health Organization Classification of Thoracic Tumors, and doctors are relatively unfamiliar with its diagnosis, treatment, and prognosis. Taking a case of SMARCA4-UT treated in Peking University First Hospital as an example, this multi-disciplinary discussion covered several hot issues on diagnosing and treating thoracic SMARCA4-UT, including histological features, immu- nohistochemical and molecular phenotype, immune checkpoint inhibitor (ICI) therapy, and pathological assessment of neoadjuvant therapy response. The patient was an older man with a long history of smoking and was admitted due to a rapidly progressing solid tumor in the lower lobe of the right lung. Histologically, tumor cells were epithelioid, undifferentiated, diffusely positive for CD34, and partially positive for SALL4.The expression of BRG1 protein encoded by SMARCA4 gene was lost in all of tumor cells, and next-generation sequencing(NGS)confirmed SMARCA4 gene mutation (c.2196T>G, p.Y732Ter). The pathological diagnosis reached as thoracic SMARCA4-UT, and the preoperative TNM stage was T1N2M0 (ⅢA). Tumor proportion score (TPS) detected by immunohistochemistry of programmed cell death 1-ligand 1 (PD-L1, clone SP263) was 2%. Tumor mutation burden (TMB) detected by NGS of 1 021 genes was 16. 3/Mb. Microsatellite detection showed the tumor was microsatellite stable (MSS). Neo-adjuvant therapy was implemented with the combined regimen of chemotherapy and ICI. Right lower lobectomy was performed through thoracoscopy after the two weeks' neoadjuvant. The pathologic assessment of lung tumor specimens after neoadjuvant therapy revealed a complete pathological response (CPR). The post-neoadjuvant tumor TNM stage was ypT0N0M0. Then, five cycles of adjuvant therapy were completed. Until October 2022, neither tumor recurrence nor metastasis was detected, and minimal residual disease (MRD) detection was negative. At present, it is believed that if BRG1 immunohistochemical staining is negative, regardless of whether SMARCA4 gene mutation is detected, it should be classified as SMARCA4-deficient tumors. SMARCA4-deficient tumors include a variety of carcinomas and sarcomas. The essential criteria for diagnosing SMARCA4-UT includes loss of BRG1 expression, speci-fic histological morphology, and exclude other common thoracic malignant tumors with SMARCA4-deficiency, such as squamous cell carcinoma, adenocarcinoma and large cell carcinoma. SMARCA4-UT is a very aggressive malignant tumor with a poor prognosis. It has almost no targeted therapy mutations, and little response to chemotherapy, but ICI is currently the only effective drug. The successful diagnosis and treatment for this case of SMARCA4-UT should enlighten significance for various kinds of SMARCA4-deficient tumors.

NUT carcinoma - An aggressive thoracic tumor.

Nuclear protein in testis (NUT) carcinoma is an extremely rare and undifferentiated malignancy characterized by the rearrangement of NUT gene (NUTM1, Nuclear Protein in Testis). NUT carcinoma is a challenging disease which is difficult to diagnose and treat. Due to its rarity, lack of experience and need of specific molecular study it can be un/misdiagnosed. Therefore, NUT carcinoma should be included in differential diagnosis of poorly differentiated/undifferentiated and rapidly progressive malignancy in children and young adults, occurring in the head, neck or thorax. We report a case of NUT carcinoma presented with pleural effusion in adulthood.

Results of screening in early and advanced thoracic malignancies in the EORTC pan-European SPECTAlung platform.

Access to a comprehensive molecular alteration screening is patchy in Europe and quality of the molecular analysis varies. SPECTAlung was created in 2015 as a pan-European screening platform for patients with thoracic malignancies. Here we report the results of almost 4 years of prospective molecular screening of patients with thoracic malignancies, in terms of quality of the program and molecular alterations identified. Patients with thoracic malignancies at any stage of disease were recruited in SPECTAlung, from June 2015 to May 2019, in 7 different countries. Molecular tumour boards were organised monthly to discuss patients' molecular and clinical profile and possible biomarker-driven treatments, including clinical trial options. FFPE material was collected and analysed for 576 patients with diagnosis of pleural, lung, or thymic malignancies. Ultimately, 539 patients were eligible (93.6%) and 528 patients were assessable (91.7%). The turn-around time for report generation and molecular tumour board was 214 days (median). Targetable molecular alterations were observed in almost 20% of cases, but treatment adaptation was low (3% of patients). SPECTAlung showed the feasibility of a pan-European screening platform. One fifth of the patients had a targetable molecular alteration. Some operational issues were discovered and adapted to improve efficiency.

The illusion of a chest wall tumor: a case-report of sternal tuberculosis.

The incidence rate of tuberculosis in developed countries is low. The most common presentation of this disease is its pulmonary form but with the increasing use of immunosuppressive drugs, extra-pulmonary tuberculosis is re-emerging. Nevertheless, sternal bone involvement is uncommon. We report the case of an eighty-three-year-old man who presented a painful sternal mass which progressed towards cutaneous ulceration. The first diagnostic hypothesis was neoplasia. The pathological and microbiological diagnosis of tuberculosis was achieved after surgical biopsy. The patient received treatment against tuberculosis for nine months enabling recovery without surgery. This case illustrates the importance of having a diagnosis prior to any kind of treatment facing any voluminous parietal thoracic lesions.  This diagnosis is made possible by surgical samples and interdisciplinary teamwork. This case underlines that tuberculosis remains a differential diagnosis that must be evoked in case of unusual bone mass.

Thorahcic SMARCA4-deficient undifferentiated tumors with ganglioneuroma and enchondroma: implications for SLC7A11 and ARID1A expression: a case report.

BACKGROUND: SWI/SNF-related, matrix-associated, actin-dependent regulator of chromatin, subfamily A, member 4-deficient thoracic sarcoma (SMARCA4-DTS) is a rare disease that has recently been described as an entity. It is characterized by an aggressive clinical course and specific genetic alterations. As an immunohistological feature, the tumors are deficient in SMARCA4 and SMARCA2 and express sex-determining region Y (SRY)-box 2 (SOX2). Occasionally, there are cases that are less frequent and difficult to distinguish from SMARCA4-deficient non-small cell lung carcinoma (SMARCA4-dNSCLC). Therefore, the 5th edition of the World Health Organization (WHO) classification describes thoracic SMARCA 4-deficient undifferentiated tumors (SMARCA4-UT). In contrast, Carney's triad is a syndrome that combines three rare soft tissue tumors: gastric leiomyosarcoma, pulmonary chondroma, and extra-adrenal paraganglioma. Protein kinase cAMP-dependent type I regulatory subunit alpha (PRKAR1A) has been proposed as the causative gene. Both diseases are valuable cases; moreover, there have been no previous reports of their coexistence. CASE PRESENTATION: A 43-year-old man visited our hospital because of respiratory distress. Computed tomography revealed a large mass measuring 55 mm in the upper lobe of the right lung and front mediastinum, with metastases in the surrounding lymph nodes. Needle biopsy was performed for diagnosis, and histological examination of the samples revealed monotonous epithelioid-like cells with loose binding and sheet-form proliferation. The tumor cells had distinct nuclei with some rhabdoid-like cells. Immunohistochemical analysis revealed that the tumor cells were positive for AE1AE3, SOX2, CD34, and p53 and negative for SMARCA4 and SMARCA2. The patient died 6 months after admission, without any treatment. Autopsy revealed ganglioneuroma and enchondroma suggestive of an incomplete Carney complex. CONCLUSION: SMARCA4-UT is a rare and recently established disease. While it is difficult to diagnose, it is necessary to distinguish undifferentiated carcinoma, large cell carcinoma, Ewing sarcoma, and epithelioid sarcoma when diagnosing tumors involving the mediastinum. Moreover, cases of SMARCA4-UT with ganglioneuroma and enchondroma are very rare. We discuss and report a case of SMARCA4-UT in which we also examined ARID1A and SLC7A11expression.

A Definitive Prognostication System for Patients With Thoracic Malignancies Diagnosed With Coronavirus Disease 2019: An Update From the TERAVOLT Registry.

INTRODUCTION: Patients with thoracic malignancies are at increased risk for mortality from coronavirus disease 2019 (COVID-19), and a large number of intertwined prognostic variables have been identified so far. METHODS: Capitalizing data from the Thoracic Cancers International COVID-19 Collaboration (TERAVOLT) registry, a global study created with the aim of describing the impact of COVID-19 in patients with thoracic malignancies, we used a clustering approach, a fast-backward step-down selection procedure, and a tree-based model to screen and optimize a broad panel of demographics and clinical COVID-19 and cancer characteristics. RESULTS: As of April 15, 2021, a total of 1491 consecutive eligible patients from 18 countries were included in the analysis. With a mean observation period of 42 days, 361 events were reported with an all-cause case fatality rate of 24.2%. The clustering procedure screened 73 covariates in 13 clusters. A further multivariable logistic regression for the association between clusters and death was performed, resulting in five clusters significantly associated with the outcome. The fast-backward step-down selection procedure then identified the following seven major determinants of death: Eastern Cooperative Oncology Group-performance status (ECOG-PS) (OR = 2.47, 1.87-3.26), neutrophil count (OR = 2.46, 1.76-3.44), serum procalcitonin (OR = 2.37, 1.64-3.43), development of pneumonia (OR = 1.95, 1.48-2.58), C-reactive protein (OR = 1.90, 1.43-2.51), tumor stage at COVID-19 diagnosis (OR = 1.97, 1.46-2.66), and age (OR = 1.71, 1.29-2.26). The receiver operating characteristic analysis for death of the selected model confirmed its diagnostic ability (area under the receiver operating curve = 0.78, 95% confidence interval: 0.75-0.81). The nomogram was able to classify the COVID-19 mortality in an interval ranging from 8% to 90%, and the tree-based model recognized ECOG-PS, neutrophil count, and c-reactive protein as the major determinants of prognosis. CONCLUSIONS: From 73 variables analyzed, seven major determinants of death have been identified. Poor ECOG-PS was found to have the strongest association with poor outcome from COVID-19. With our analysis, we provide clinicians with a definitive prognostication system to help determine the risk of mortality for patients with thoracic malignancies and COVID-19.

Intraoperative Cytology for Video-Assisted Thoracic Surgery: A Quality Improvement Analysis.

BACKGROUND: Frozen section is a standard of care procedure during thoracic surgery when an immediate diagnosis is needed. An alternative procedure is intraoperative cytology. Video-assisted thoracic surgery is currently widely used for thoracic surgical procedures. The aim of this study was to assess intraoperative cytology together with frozen section for accuracy, turnaround time, and total response time during video-assisted thoracic surgery. METHODS: We included patients having video-assisted thoracic surgery between August 2018 and February 2019 at our institution. A cytopathologist and a surgical pathologist independently performed intraoperative cytology and frozen sections, respectively. Final histologic diagnosis was the reference standard. Intraoperative cytology, frozen section turnaround, and total response times were analyzed. RESULTS: A total of 52 specimens from 27 patients were included. The intraoperative cytology correlated with final histology in 98% of cases. Frozen section correlated with final histology in 100% of cases. Intraoperative cytology turnaround and total response times were equal (mean, 4.35 minutes; range, 2-15 minutes). Mean frozen section turnaround and response times were 26.2 minutes (range, 9-61 minutes) and 36.7 minutes (range, 16-90 minutes), respectively. We found a statistically significant difference between intraoperative cytology and frozen section turnaround time and total response times (P < .001). CONCLUSIONS: This study highlights that intraoperative cytology could be as accurate as frozen section and considerably faster during video-assisted thoracic surgery (P < .001). Total response time could potentially be used as a quality metric for video-assisted thoracic surgery.

Novel fusion sarcomas including targetable NTRK and ALK.

BACKGROUND: Challenging emerging entities with distinctive molecular signatures may benefit from algorithms for diagnostic work-up. METHODS: Fusion sarcomas (2020-2021, during pandemic) were diagnosed by clinicoradiology, morphology, phenotype, and next-generation sequencing (NGS). RESULTS: Six fusion sarcomas in two males and four females involved the chest-wall, neck, or extremities; ages ranged 2-73, median 18 years. Sizes ranged 5.3-25.0, median 9.1 cm. These include high grade 1) TPR-NTRK1 of proximal femur with a larger rounded soft tissue mass, previously considered osteosarcoma yet without convincing tumor matrix. A pathologic fracture necessitated emergency hemipelvectomy (NED) and 2) novel KANK1-NTRK2 sarcoma of bone and soft tissue with spindled pleomorphic to epithelioid features (AWD metastases). 3) Novel ERC1-ALK unaligned fusion, a low grade infiltrative deep soft tissue hand sarcoma with prominent-vascularity, myopericytoid/lipofibromatosis-like ovoid cells, and collagenized stroma, was successfully treated with ALK-inhibitor (Crizotinib), avoiding amputation. These NTRK and ALK tumors variably express S100 and CD34 and were negative for SOX10. 4) and 5) CIC-DUX4 round cell tumors (rapid metastases/demise), one with COVID superinfection, were previously treated as Ewing sarcoma. These demonstrated mild pleomorphism and necrosis, variable myxoid change and CD99 reactivity, and a distinctive dot-like-Golgi WT1 immunostaining pattern. 6) A chest wall/thoracic round cell sarcoma, focal CD34/ keratins/CK7, revealed nuclear-STAT6, STAT6-NAB2 by NGS, confirming malignant solitary fibrous tumor, intermediate-risk-stratification (AWD metastases). CONCLUSIONS: Recent fusion sarcomas include new KANK1-NTRK2 and ERC1-ALK, the latter successfully treated by targeted-therapy. ALK/NTRK fusion partners TPR and KANK1 suggest unusual high-grade morphology/behavior. Clinicoradiologic, morphologic, and phenotypic algorithms can prompt molecular-targeted immunostains or NGS for final classification and promising inhibitor therapy.

Effect of a comprehensive deep-learning model on the accuracy of chest x-ray interpretation by radiologists: a retrospective, multireader multicase study.

BACKGROUND: Chest x-rays are widely used in clinical practice; however, interpretation can be hindered by human error and a lack of experienced thoracic radiologists. Deep learning has the potential to improve the accuracy of chest x-ray interpretation. We therefore aimed to assess the accuracy of radiologists with and without the assistance of a deep-learning model. METHODS: In this retrospective study, a deep-learning model was trained on 821 681 images (284 649 patients) from five data sets from Australia, Europe, and the USA. 2568 enriched chest x-ray cases from adult patients (≥16 years) who had at least one frontal chest x-ray were included in the test dataset; cases were representative of inpatient, outpatient, and emergency settings. 20 radiologists reviewed cases with and without the assistance of the deep-learning model with a 3-month washout period. We assessed the change in accuracy of chest x-ray interpretation across 127 clinical findings when the deep-learning model was used as a decision support by calculating area under the receiver operating characteristic curve (AUC) for each radiologist with and without the deep-learning model. We also compared AUCs for the model alone with those of unassisted radiologists. If the lower bound of the adjusted 95% CI of the difference in AUC between the model and the unassisted radiologists was more than -0·05, the model was considered to be non-inferior for that finding. If the lower bound exceeded 0, the model was considered to be superior. FINDINGS: Unassisted radiologists had a macroaveraged AUC of 0·713 (95% CI 0·645-0·785) across the 127 clinical findings, compared with 0·808 (0·763-0·839) when assisted by the model. The deep-learning model statistically significantly improved the classification accuracy of radiologists for 102 (80%) of 127 clinical findings, was statistically non-inferior for 19 (15%) findings, and no findings showed a decrease in accuracy when radiologists used the deep-learning model. Unassisted radiologists had a macroaveraged mean AUC of 0·713 (0·645-0·785) across all findings, compared with 0·957 (0·954-0·959) for the model alone. Model classification alone was significantly more accurate than unassisted radiologists for 117 (94%) of 124 clinical findings predicted by the model and was non-inferior to unassisted radiologists for all other clinical findings. INTERPRETATION: This study shows the potential of a comprehensive deep-learning model to improve chest x-ray interpretation across a large breadth of clinical practice. FUNDING: Annalise.ai.

Predictive online 3D target tracking with population-based generative networks for image-guided radiotherapy.

PURPOSE: Respiratory motion of thoracic organs poses a severe challenge for the administration of image-guided radiotherapy treatments. Providing online and up-to-date volumetric information during free breathing can improve target tracking, ultimately increasing treatment efficiency and reducing toxicity to surrounding healthy tissue. In this work, a novel population-based generative network is proposed to address the problem of 3D target location prediction from 2D image-based surrogates during radiotherapy, thus enabling out-of-plane tracking of treatment targets using images acquired in real time. METHODS: The proposed model is trained to simultaneously create a low-dimensional manifold representation of 3D non-rigid deformations and to predict, ahead of time, the motion of the treatment target. The predictive capabilities of the model allow correcting target location errors that can arise due to system latency, using only a baseline volume of the patient anatomy. Importantly, the method does not require supervised information such as ground-truth registration fields, organ segmentation, or anatomical landmarks. RESULTS: The proposed architecture was evaluated on both free-breathing 4D MRI and ultrasound datasets. Potential challenges present in a realistic therapy, like different acquisition protocols, were taken into account by using an independent hold-out test set. Our approach enables 3D target tracking from single-view slices with a mean landmark error of 1.8 mm, 2.4 mm and 5.2 mm in volunteer MRI, patient MRI and US datasets, respectively, without requiring any prior subject-specific 4D acquisition. CONCLUSIONS: This model presents several advantages over state-of-the-art approaches. Namely, it benefits from an explainable latent space with explicit respiratory phase discrimination. Thanks to the strong generalization capabilities of neural networks, it does not require establishing inter-subject correspondences. Once trained, it can be quickly deployed with an inference time of only 8 ms. The results show the capability of the network to predict future anatomical changes and track tumors in real time, yielding statistically significant improvements over related methods.

SMARCA4-deficient thoracic sarcoma revealed by metastasis to the small intestine: a diagnostic dilemma.

SMARCA4-deficient thoracic sarcoma (SMARCA4-DTS) is a recently identified aggressive subtype of sarcoma. We present the case of a 44-year-old man who underwent a surgery for a perforated small intestine. Compued tomography scan revealed a tissular mediastino-pulmonary mass.Histopathological examination of the intestinal mass shown a malignant tumour with a typical epithelioid and rhabdoid cells, numerous mitoses and large necrosis. A large panel of immunohistochemistry revealed loss of SMARCA4 and SMARCA2 and allowed the diagnosis of SMARCA4-DTS. It is important to consider SMARCA4-deficient thoracic sarcoma in the differential diagnosis of tumours showing suggestive morphologic features in patients of all ages, especially in the case of metastasis associated with thoracic mass.