Citations

Objective CHAMPION‐NMOSD (NCT04201262) is a phase 3, open‐label, externally controlled interventional study evaluating the efficacy and safety of the terminal complement inhibitor ravulizumab in adult patients with anti–aquaporin‐4 antibody–positive (AQP4+) neuromyelitis optica spectrum disorder (NMOSD). Ravulizumab binds the same complement component 5 epitope as the approved therapeutic eculizumab but has a longer half‐life, enabling an extended dosing interval (8 vs 2 weeks). Methods The availability of eculizumab precluded the use of a concurrent placebo control in CHAMPION‐NMOSD; consequently, the placebo group of the eculizumab phase 3 trial PREVENT (n = 47) was used as an external comparator. Patients received weight‐based intravenous ravulizumab on day 1 and maintenance doses on day 15, then once every 8 weeks. The primary endpoint was time to first adjudicated on‐trial relapse. Results The primary endpoint was met; no patients taking ravulizumab (n = 58) had an adjudicated relapse (during 84.0 patient‐years of treatment) versus 20 patients with adjudicated relapses in the placebo group of PREVENT (during 46.9 patient‐years; relapse risk reduction = 98.6%, 95% confidence interval = 89.7%–100.0%, p < 0.0001). Median (range) study period follow‐up time was 73.5 (11.0–117.7) weeks for ravulizumab. Most treatment‐emergent adverse events were mild/moderate; no deaths were reported. Two patients taking ravulizumab experienced meningococcal infections. Both recovered with no sequelae; one continued ravulizumab treatment. Interpretation Ravulizumab significantly reduced relapse risk in patients with AQP4+ NMOSD, with a safety profile consistent with those of eculizumab and ravulizumab across all approved indications. ANN NEUROL 2023;93:1053–1068

AbstractBackgroundRecent advancements in Deep Learning (DL) methodologies have led to state‐of‐the‐art performance in a wide range of applications especially in object recognition, classification, and segmentation of medical images. However, training modern DL models requires a large amount of computation and long training times due to the complex nature of network structures and the large number of training datasets involved. Moreover, it is an intensive, repetitive manual process to select the optimized configuration of hyperparameters for a given DL network.PurposeIn this study, we present a novel approach to accelerate the training time of DL models via the progressive feeding of training datasets based on similarity measures for medical image segmentation. We term this approach Progressive Deep Learning (PDL).MethodsThe two‐stage PDL approach was tested on the auto‐segmentation task for two imaging modalities: CT and MRI. The training datasets were ranked according to similarity measures between each sample based on Mean Square Error (MSE), Peak Signal‐to‐Noise Ratio (PSNR), Structural Similarity Index (SSIM), and the Universal Quality Image Index (UQI) values. At the start of the training process, a relatively coarse sampling of training datasets with higher ranks was used to optimize the hyperparameters of the DL network. Following this, the samples with higher ranks were used in step 1 to yield accelerated loss minimization in early training epochs and the total dataset was added in step 2 for the remainder of training.ResultsOur results demonstrate that the PDL approach can reduce the training time by nearly half (∼49%) and can predict segmentations (CT U‐net/DenseNet dice coefficient: 0.9506/0.9508, MR U‐net/DenseNet dice coefficient: 0.9508/0.9510) without major statistical difference (Wilcoxon signed‐rank test) compared to the conventional DL approach. The total training times with a fixed cutoff at 0.95 DSC for the CT dataset using DenseNet and U‐Net architectures, respectively, were 17 h, 20 min and 4 h, 45 min in the conventional case compared to 8 h, 45 min and 2 h, 20 min with PDL. For the MRI dataset, the total training times using the same architectures were 2 h, 54 min and 52 min in the conventional case and 1 h, 14 min and 25 min with PDL.ConclusionThe proposed PDL training approach offers the ability to substantially reduce the training time for medical image segmentation while maintaining the performance achieved in the conventional case.

Abstract Aims To investigate whether left arterial reservoir strain (LASr) could predict new-onset atrial fibrillation (NOAF) in patients with light-chain-type cardiac amyloidosis (ALCA). Methods and results This study enrolled 427 patients with CA from two tertiary centres between 2005 and 2019. LASr was measured using a vendor-independent analysis programme. The primary outcome was NOAF. A total of 287 patients with ALCA were included [median age 63.0 (56.0–70.0) years, 53.3% male]. The median LASr was 13.9% (10.5–20.8%). During the median follow-up of 0.85 years, AF occurred in 34 patients (11.8%). In the receiver operating characteristics curve analysis, the optimal cut-off of LASr for predicting NOAF was 14.4%. Patients with LASr ≤14.4% had a higher risk of NOAF than those with LASr &gt;14.4% (18.1% vs. 5.1%, P &lt; 0.010). In the multivariate analysis adjusting for confounding factors, including left arterial volume index and left ventricular global longitudinal strain (LV-GLS), higher LASr (%) was independently associated with lower risk for NOAF [adjusted hazard ratio (aHR): 0.936, 95% confidence interval (95% CI): 0.879–0.997, P = 0.039]. Furthermore, LASr ≤14.4% was an independent predictor for NOAF (aHR: 3.370, 95% CI: 1.337–8.492, P = 0.010). This remained true after accounting for all-cause death as a competing risk. Compared with Model 1 (LV-GLS) and Model 2 (LV-GLS plus LAVI), Model 3, including LASr showed a better reclassification ability for predicting NOAF (net reclassification index = 0.735, P &lt; 0.001 compared with Model 1; net reclassification index = 0.514, P = 0.003 compared with Model 2). Conclusion LASr was an independent predictor of NOAF in patients with ALCA.

Abstract The tumor microenvironment (TME) plays a crucial role in mediating the tumor immune response, thereby affecting patient outcomes in follicular lymphoma (FL). Using CyTOF, we analyzed a cohort of 82 FL patients with biopsy specimens collected before the first treatment regimen was administered. Our results showed that there was a difference in the T-cell phenotype in patients with various TMEs. Patients with TMEs rich in monocyte/macrophages/NK cells had more CD8+ cells and tended to have fewer TN, but more TEMRA and TEXH cells. Patients enriched with T cells exhibited a T-cell phenotype skewed toward to TCM and patients with B cell dominance tended to have significant higher number of TFH cells. T-cell phenotypes with more terminally differentiated, senescent and exhausted cells were highly represented in patients with less T cells when compared to patients with more T cells. We observed that CD57+ TFH cells was significantly more abundant in patients who had disease progression than patients who had complete response to therapy, suggesting a role for CD57+ TFH cells in promoting malignant cell growth in FL. Consistent with this finding, we observed that increased numbers of CD57+ TFH cells correlated with an inferior survival in FL. Using CITE-seq technology, we found that CD57+ TFH cells exhibited a substantially different transcriptome from CD57− TFH cells. Genes that were differentially upregulated in CD57+ TFH cells when compared to CD57− TFH cells included genes involved in cell survival, compromised inflammatory response, and metabolic activation. Taken together, our results indicate different tumor microenvironments among patient groups that is associated with variable T-cell phenotypes.

Abstract T cell Ig and ITIM domain (TIGIT) is an immune checkpoint molecule and its ligation delivers an inhibitory signal to T cells that negatively regulates anti-tumor responses. However, the expression and biological relevance of TIGIT in B-cell non-Hodgkin lymphoma (NHL) is completely unknown. To phenotypically characterize TIGIT+ T cell subsets, we profiled TIGIT-expressing CD4+ and CD8+ T cells from biopsy specimens of B-cell NHL. TIGIT expression was not detectable on T cells from peripheral blood, however, TIGIT is constitutively expressed on CD4+ T cells from biopsy specimens of B-cell NHL with a median of 44.8% (range: 25.9–62.2, n=8) of CD4+ T cells expressing TIGIT. By profiling the CD4+TIGIT+ T cells, we observed that CD25 was highly expressed on intratumoral CD4+TIGIT+ T cells. Intracellular staining revealed that Foxp3 was expressed by intratumoral CD4+TIGIT+ T cells. These intratumoral CD4+TIGIT+ T cells expressed high level of chemokine receptor CCR4 and were absent of IL-7 receptor-α (CD127). Functionally, CD4+TIGIT+ T cells displayed reduced cytokine production, as the number of IFN-γ- and TNF-α-producing cells was lower in the TIGIT+ population than in TIGIT− T cells. While TIGIT expression can be induced by TCR activation, treatment of T cells with IL-2 and TGF-β inhibited TCR-mediated TIGIT induction. Furthermore, lymphoma B cells were involved in TIGIT induction as in vitro depletion of lymphoma B cells altered upregulation of TIGIT on CD4+ T cells in B-cell NHL. Taken together, these results indicated that constitutive expression of TIGIT defines a CD4+ population with similar phenotype to regulatory T cells. Inhibition of TIGIT signaling may be an additional mechanism to prevent T-cell suppression in B-cell NHL

Abstract The low-affinity IgG Fc receptor, FcγRII (CD32), mediates various effector functions of lymphoid and myeloid cells and is the major IgG Fc receptor expressed by human eosinophils. We investigated whether FcγRII regulates both cell survival and death of human eosinophils. When cultured in vitro without growth factors, most eosinophils undergo apoptosis within 96 h. Ligation of FcγRII by anti-CD32 mAb in solution inhibited eosinophil apoptosis and prolonged survival in the absence of growth factors. Cross-linking of human IgG bound to FcγRII by anti-human IgG Ab or of unoccupied FcγRII by aggregated human IgG also prolonged eosinophil survival. The enhanced survival with anti-CD32 mAb was inhibited by anti-granulocyte-macrophage-CSF (GM-CSF) mAb, suggesting that autocrine production of GM-CSF by eosinophils mediated survival. In fact, mRNA for GM-CSF was detected in eosinophils cultured with anti-CD32 mAb. In contrast to mAb or ligands in solution, anti-CD32 mAb or human IgG, when immobilized onto tissue culture plates, facilitated eosinophil cell death even in the presence of IL-5. Cell death induced by these immobilized ligands was accompanied by DNA fragmentation and was inhibited when eosinophil β2 integrin was blocked by anti-CD18 mAb, suggesting that β2 integrins play a key role in initiating eosinophil apoptosis. Thus, FcγRII may pivotally regulate both survival and death of eosinophils, depending on the manner of receptor ligation and β2 integrin involvement. Moreover, the FcγRII could provide a novel mechanism to control the number of eosinophils at inflammation sites in human diseases.

Abstract Background: The antibiotic resistance of pathogenic microbes due to the overuse of antibiotics is a growing threat to the public health and can eventually reduce the therapeutic effects of antibiotics. Hydrogel-based systems for local delivery of antibiotics can be a solution to this problem. Metal-organic frameworks (MOFs) have received considerable interest for use as antimicrobial materials due to their high porosity, large surface area to volume, diverse structures, and tunable topologies. Methods: In this study, we developed MOF-encapsulated and dual (photo & chemical)-crosslinkable alginate hydrogel as an antimicrobial scaffold. We synthesized antimicrobial Cu(AZPY)-MOF using solvothermal reaction and encapsulated them in an alginate-based hydrogel via visible light photo-crosslinking and calcium ion-induced chemical-crosslinking processes. Results: Powder X-ray diffraction, Fourier-transform infrared spectroscopy, thermogravimetric analysis, and scanning electron microscopy confirmed the successful fabrication of Cu(AZPY)-MOF encapsulated alginate hydrogel. The mechanical strength of the dual-crosslinked hydrogels was better than that of the photo-crosslinked hydrogel. Encapsulation of Cu(AZPY)-MOF in an alginate hydrogel resulted in a further increase in the mechanical strength of the hydrogel owing to the additional interaction of alginate with Cu-MOF. Moreover, the Cu(AZPY)-MOF encapsulated alginate hydrogel exhibited excellent antibacterial and antifungal properties against two bacterial strains (i.e., MRSA and S. mutans) and one fungal strain (C. albicans) as well as negligible cytotoxicity toward MEFs. Conclusions: The antimicrobial agents encapsulated dual-crosslinkable alginate hydrogel developed in this study is a promising candidate for use in tissue engineering and biomedical field.

AbstractBackgroundThe growing adoption of magnetic resonance imaging (MRI)‐guided radiation therapy (RT) platforms and a focus on MRI‐only RT workflows have brought the technical challenge of synthetic computed tomography (sCT) reconstruction to the forefront. Unpaired‐data deep learning‐based approaches to the problem offer the attractive characteristic of not requiring paired training data, but the gap between paired‐ and unpaired‐data results can be limiting.PurposeWe present two distinct approaches aimed at improving unpaired‐data sCT reconstruction results: a cascade ensemble that combines multiple models and a personalized training strategy originally designed for the paired‐data setting.MethodsComparisons are made between the following models: (1) the paired‐data fully convolutional DenseNet (FCDN), (2) the FCDN with the Intentional Deep Overfit Learning (IDOL) personalized training strategy, (3) the unpaired‐data CycleGAN, (4) the CycleGAN with the IDOL training strategy, and (5) the CycleGAN as an intermediate model in a cascade ensemble approach. Evaluation of the various models over 25 total patients is carried out using a five‐fold cross‐validation scheme, with the patient‐specific IDOL models being trained for the five patients of fold 3, chosen at random.ResultsIn both the paired‐ and unpaired‐data settings, adopting the IDOL training strategy led to improvements in the mean absolute error (MAE) between true CT images and sCT outputs within the body contour (mean improvement, paired‐ and unpaired‐data approaches, respectively: 38%, 9%) and in regions of bone (52%, 5%), the peak signal‐to‐noise ratio (PSNR; 15%, 7%), and the structural similarity index (SSIM; 6%, <1%). The ensemble approach offered additional benefits over the IDOL approach in all three metrics (mean improvement over unpaired‐data approach in fold 3; MAE: 20%; bone MAE: 16%; PSNR: 10%; SSIM: 2%), and differences in body MAE between the ensemble approach and the paired‐data approach are statistically insignificant.ConclusionsWe have demonstrated that both a cascade ensemble approach and a personalized training strategy designed initially for the paired‐data setting offer significant improvements in image quality metrics for the unpaired‐data sCT reconstruction task. Closing the gap between paired‐ and unpaired‐data approaches is a step toward fully enabling these powerful and attractive unpaired‐data frameworks.