BACKGROUND AND PURPOSE:Early assessment of treatment response is critical in patients with glioblastomas. A combination of DTI and DSC perfusion imaging parameters was evaluated to distinguish glioblastomas with true progression from mixed response and pseudoprogression.

Background: Differentiating glioblastoma, brain metastasis, and central nervous system lymphoma (CNSL) on conventional magnetic resonance imaging (MRI) can present a diagnostic dilemma due to the potential for overlapping imaging features. We investigate whether machine learning evaluation of multimodal MRI can reliably differentiate these entities. Methods: Preoperative brain MRI including diffusion weighted imaging (DWI), dynamic contrast enhanced (DCE), and dynamic susceptibility contrast (DSC) perfusion in patients with glioblastoma, lymphoma, or metastasis were retrospectively reviewed. Perfusion maps (rCBV, rCBF), permeability maps (K-trans, Kep, Vp, Ve), ADC, T1C+ and T2/FLAIR images were coregistered and two separate volumes of interest (VOIs) were obtained from the enhancing tumor and non-enhancing T2 hyperintense (NET2) regions. The tumor volumes obtained from these VOIs were utilized for supervised training of support vector classifier (SVC) and multilayer perceptron (MLP) models. Validation of the trained models was performed on unlabeled cases using the leave-one-subject-out method. Head-to-head and multiclass models were created. Accuracies of the multiclass models were compared against two human interpreters reviewing conventional and diffusion-weighted MR images. Results: Twenty-six patients enrolled with histopathologically-proven glioblastoma (n=9), metastasis (n=9), and CNS lymphoma (n=8) were included. The trained multiclass ML models discriminated the three pathologic classes with a maximum accuracy of 69.2% accuracy (18 out of 26; kappa 0.540, P=0.01) using an MLP trained with the VpNET2 tumor volumes. Human readers achieved 65.4% (17 out of 26) and 80.8% (21 out of 26) accuracies, respectively. Using the MLP VpNET2 model as a computer-aided diagnosis (CADx) for cases in which the human reviewers disagreed with each other on the diagnosis resulted in correct diagnoses in 5 (19.2%) additional cases. Conclusions: Our trained multiclass MLP using VpNET2 can differentiate glioblastoma, brain metastasis, and CNS lymphoma with modest diagnostic accuracy and provides approximately 19% increase in diagnostic yield when added to routine human interpretation.

Gastric carcinosarcoma is an unusual tumor and its occurrence in the gastric stump is extremely rare. A report is presented here of a unique case of gastric stump carcinosarcoma with rhabdomyosarcomatous differentiation in a 74-year-old man. The patient had undergone partial gastrectomy with gastrojejunostomy (Billroth II method) 30 years previously. The tumor had both adenocarcinoma and sarcoma components, and an immunohistochemical study suggested a focal transition between these components. The main sarcomatous components showed fibrosarcomatous features with a scattered distribution of rounded tumor cells, whose rhabdomyosarcomatous differentiation was immunohistochemically determined. Ultrastructural examination supported the rhabdomyosarcomatous natures. Experience with the present tumor indicates that carcinosarcoma with rhabdomyosarcomatous differentiation can occur in the gastric stump and that this disease is capable of aggressive behavior.

Intravascular large B‐cell lymphoma (IVLBCL) is a distinct disease, but the neoplastic PD‐L1 expression on tumor cells may vary among cases. We evaluated 10 IVLBCL autopsy cases for neoplastic PD‐L1 expression, and had positive results in two cases. In one case, neoplastic PD‐L1 expression (SP142, 28‐8, and E1J2J clones) was dependent on the organ and anatomical site (capillaries vs. vessels) of the tumor tissue. Neoplastic PD‐L1 expression was found in tumor cells located in capillaries in the central nervous system, pituitary gland, kidneys, lung, and gastrointestinal tract; sinuses/sinusoids of the spleen, liver, bone marrow, and lymph nodes; and an extravascular location. However, this expression was not detected in tumor cells located in the adrenal gland, thyroid gland, pancreas, ovaries, uterus, pleura, and small or larger‐sized vessels of the lung. The other case showed constant neoplastic PD‐L1 expression on the tumor cells, and in addition to the affected organs, capillaries, and vessels with two anti‐PD‐L1 antibodies (28‐8 and E1J2J, but not SP142). The divergence and heterogeneity of neoplastic PD‐L1 expression were clearly demonstrated in our cases. To the best of our knowledge, this is the first description of divergent neoplastic PD‐L1 expression among the affected organs and anatomical sites in IVLBCL.

BackgroundPerihematomal edema (PHE) volume correlates with intracerebral hemorrhage (ICH) volume and is associated with functional outcome. Minimally invasive surgery (MIS) for ICH decreases clot burden and PHE. MIS may therefore alter the time course of PHE, mitigating a critical source of secondary injury.ObjectiveTo describe a new method for the quantitative measurement of cerebral edema surrounding the evacuated hematoma cavity, termed pericavity edema (PCE), and obtain details of its time course following MIS for ICH.MethodsThe study included 48 consecutive patients presenting with ICH who underwent MIS evacuation. Preoperative and postoperative CT scans were assessed by two independent raters. Hematoma, edema, cavity, and pneumocephalus volumes were calculated using semi-automatic, threshold-guided volume segmentation software (AnalyzePro). Follow-up CT scans at variable delayed time points were available for 36 patients and were used to describe the time course of PCE.ResultsMean preoperative, postoperative, and delayed PCE were 21.0 mL (SD 15.5), 18.6 mL (SD 11.4), and 18.4 mL (SD 15.5), respectively. The percentage of ICH evacuated correlated significantly with a decrease in postoperative PCE (r=−0.46, p<0.01). Linear regression analysis revealed a significant relation between preoperative hematoma volume and both postoperative PCE (p<0.001) and postoperative relative PCE (p<0.001). The mean peak PCE was 26.4 mL (SD 15.6) and occurred at 6.5 days (SD 4.8) post-ictus. The 2-week postoperative time course of relative PCE did not fluctuate, suggesting stability in edema during the perioperative period surrounding evacuation and up to 2 weeks after the initial bleed.ConclusionsWe present a detailed and accurate method for measuring PCE volume with semi-automatic, threshold-guided segmentation software in the postoperative patient with ICH. Decrease in PCE after MIS evacuation correlated with evacuation percentage, and relative PCE remained stable after minimally invasive endoscopic ICH evacuation.