Improving Radiographic Fracture Recognition Performance and Efficiency Using Artificial Intelligence.

Background Missed fractures are a common cause of diagnostic discrepancy between initial radiographic interpretation and the final read by board-certified radiologists. Purpose To assess the effect of assistance by artificial intelligence (AI) on diagnostic performances of physicians for fractures on radiographs. Materials and Methods This retrospective diagnostic study used the multi-reader, multi-case methodology based on an external multicenter data set of 480 examinations with at least 60 examinations per body region (foot and ankle, knee and leg, hip and pelvis, hand and wrist, elbow and arm, shoulder and clavicle, rib cage, and thoracolumbar spine) between July 2020 and January 2021. Fracture prevalence was set at 50%. The ground truth was determined by two musculoskeletal radiologists, with discrepancies solved by a third. Twenty-four readers (radiologists, orthopedists, emergency physicians, physician assistants, rheumatologists, family physicians) were presented the whole validation data set (n = 480), with and without AI assistance, with a 1-month minimum washout period. The primary analysis had to demonstrate superiority of sensitivity per patient and the noninferiority of specificity per patient at -3% margin with AI aid. Stand-alone AI performance was also assessed using receiver operating characteristic curves. Results A total of 480 patients were included (mean age, 59 years ± 16 [standard deviation]; 327 women). The sensitivity per patient was 10.4% higher (95% CI: 6.9, 13.9; P < .001 for superiority) with AI aid (4331 of 5760 readings, 75.2%) than without AI (3732 of 5760 readings, 64.8%). The specificity per patient with AI aid (5504 of 5760 readings, 95.6%) was noninferior to that without AI aid (5217 of 5760 readings, 90.6%), with a difference of +5.0% (95% CI: +2.0, +8.0; P = .001 for noninferiority). AI shortened the average reading time by 6.3 seconds per examination (95% CI: -12.5, -0.1; P = .046). The sensitivity by patient gain was significant in all regions (+8.0% to +16.2%; P < .05) but shoulder and clavicle and spine (+4.2% and +2.6%; P = .12 and .52). Conclusion AI assistance improved the sensitivity and may even improve the specificity of fracture detection by radiologists and nonradiologists, without lengthening reading time. Published under a CC BY 4.0 license. Online supplemental material is available for this article. See also the editorial by Link and Pedoia in this issue.

Notch receptors and Smad3 signaling cooperate in the induction of interleukin-9-producing T cells.

Interleukin 9 (IL-9) is a pleiotropic cytokine that can regulate autoimmune responses by enhancing regulatory CD4(+)FoxP3(+) T regulatory (Treg) cell survival and T helper 17 (Th17) cell proliferation. Here, we analyzed the costimulatory requirements for the induction of Th9 cells, and demonstrated that Notch pathway cooperated with TGF-ß signaling to induce IL-9. Conditional ablation of Notch1 and Notch2 receptors inhibited the development of Th9 cells. Notch1 intracellular domain (NICD1) recruited Smad3, downstream of TGF-ß cytokine signaling, and together with recombining binding protein (RBP)-Jκ bound the Il9 promoter and induced its transactivation. In experimental autoimmune encephalomyelitis (EAE), Jagged2 ligation regulated clinical disease in an IL-9-dependent fashion. Signaling through Jagged2 expanded Treg cells and suppressed EAE when administered before antigen immunization, but worsened EAE when administered concurrently with immunization by favoring Th17 cell expansion. We propose that Notch and Smad3 cooperate to induce IL-9 and participate in regulating the immune response.

Jagged2-signaling promotes IL-6-dependent transplant rejection.

The Notch pathway is an important intercellular signaling pathway that plays a major role in controlling cell fate. Accumulating evidence indicates that Notch and its ligands present on antigen-presenting cells might be important mediators of T helper cell differentiation. In this study, we investigated the role of Jagged2 in murine cardiac transplantation by using a signaling Jagged2 mAb (Jag2) that activates recombinant signal-binding protein-Jκ. While administration of Jag2 mAb had little effect on graft survival in the fully allogeneic mismatched model BALB/c→B6, it hastened rejection in CD28-deficient recipients. Similarly, Jag2 precipitated rejection in the bm12→B6 model. In this MHC class II-mismatched model, allografts spontaneously survive for >56 days due to the emergence of Treg cells that inhibit the expansion of alloreactive T cells. The accelerated rejection was associated with upregulation of Th2 cytokines and proinflammatory cytokine IL-6, despite expansion of Treg cells. Incubation of Treg cells with recombinant IL-6 abrogated their inhibitory effects in vitro. Furthermore, neutralization of IL-6 in vivo protected Jag2-treated recipients from rejection and Jagged2 signaling was unable to further accelerate rejection in the absence of Treg cells. Our findings therefore suggest that Jagged2 signaling can affect graft acceptance by upregulation of IL-6 and consequent resistance to Treg-cell suppression.

Notch ligand delta-like 4 blockade alleviates experimental autoimmune encephalomyelitis by promoting regulatory T cell development.

Notch signaling pathway plays an important role in T cell differentiation. Delta-like ligand (Dll)4, one of five known Notch ligands, has been implicated in regulating Th2 cell differentiation in animal models of human diseases. However, the role of Dll4 in Th1/Th17-mediated autoimmune diseases remains largely unknown. Using an anti-Dll4 blocking mAb, we show that neutralizing Dll4 during the induction phase of experimental autoimmune encephalomyelitis in C57BL/6 mice significantly increased the pool of CD4(+)Foxp3(+) regulatory T cells (Treg) in the periphery and in the CNS, and decreased the severity of clinical disease and CNS inflammation. Dll4 blockade promoted induction of myelin-specific Th2/Treg immune responses and impaired Th1/Th17 responses compared with IgG-treated mice. In vitro, we show that signaling with recombinant Dll4 inhibits the TGF-ß-induced Treg development, and inhibits Janus kinase 3-induced STAT5 phosphorylation, a transcription factor known to play a key role in Foxp3 expression and maintenance. Depletion of natural Treg using anti-CD25 Ab reversed the protective effects of anti-Dll4 Ab. These findings outline a novel role for Dll4-Notch signaling in regulating Treg development in EAE, making it an encouraging target for Treg-mediated immunotherapy in autoimmune diseases, such as multiple sclerosis.

Towards a novel clinical outcome assessment for systemic lupus erythematosus: first outcomes of an international taskforce.

Systemic lupus erythematosus (SLE) is a disease of high unmet therapeutic need. The challenge of accurately measuring clinically meaningful responses to treatment has hindered progress towards positive outcomes in SLE trials, impeding the approval of potential new therapies. Current primary end points used in SLE trials are based on legacy disease activity measures that were neither specifically designed for the clinical trial context, nor developed according to contemporary recommendations for clinical outcome assessments (COAs), such as that substantial patient input should be incorporated into their design. The Treatment Response Measure for SLE (TRM-SLE) Taskforce is a global collaboration of SLE clinician-academics, patients and patient representatives, industry partners and regulatory experts, established to realize the goal of developing a new COA for SLE clinical trials. The aim of this project is a novel COA designed specifically to measure treatment effects that are clinically meaningful to patients and clinicians, and intended for implementation in a trial end point that supports regulatory approval of novel therapeutic agents in SLE. This Consensus Statement reports the first outcomes of the TRM-SLE project, including a structured process for TRM-SLE development.

Reproducibility and accuracy of quantitative myocardial blood flow assessment with (82)Rb PET: comparison with (13)N-ammonia PET.

UNLABELLED: (82)Rb cardiac PET allows the assessment of myocardial perfusion with a column generator in clinics that lack a cyclotron. There is evidence that the quantitation of myocardial blood flow (MBF) and coronary flow reserve (CFR) with dynamic (82)Rb PET is feasible. The objectives of this study were to determine the accuracy and reproducibility of MBF estimates from dynamic (82)Rb PET by using our methodology for generalized factor analysis (generalized factor analysis of dynamic sequences [GFADS]) and compartment analysis. METHODS: Reproducibility was evaluated in 22 subjects undergoing dynamic rest and dipyridamole stress (82)Rb PET studies at a 2-wk interval. The inter- and intraobserver variability of MBF quantitation with dynamic (82)Rb PET was assessed with 4 repeated estimations by each of 4 observers. Accuracy was evaluated in 20 subjects undergoing dynamic rest and dipyridamole stress PET studies with (82)Rb and (13)N-ammonia, respectively. The left ventricular and right ventricular blood pool and left ventricular tissue time-activity curves were estimated by GFADS. MBF was estimated by fitting the blood pool and tissue time-activity curves to a 2-compartment kinetic model for (82)Rb and to a 3-compartment model for (13)N-ammonia. CFR was estimated as the ratio of peak MBF to baseline MBF. RESULTS: The reproducibility of the MBF estimates in repeated (82)Rb studies was very good at rest and during peak stress (R(2)= 0.935), as was the reproducibility of the CFR estimates (R(2) = 0.841). The slope of the correlation line was very close to one for the estimation of MBF (0.986) and CFR (0.960) in repeated (82)Rb studies. The intraobserver reliability was less than 3% for the estimation of MBF at rest and during peak stress as well as for the estimation of CFR. The interobserver reliabilities were 0.950 at rest and 0.975 at peak stress. The correlation between myocardial flow estimates obtained at rest and those obtained during peak stress in (82)Rb and (13)N-ammonia studies was very good (R(2) = 0.857). Bland-Altman plots comparing CFR estimated with (82)Rb and CFR estimated with (13)N-ammonia revealed an underestimation of CFR with (82)Rb compared with (13)N-ammonia; the underestimation was within +/-1.96 SD. CONCLUSION: MBF quantitation with GFADS and dynamic (82)Rb PET demonstrated excellent reproducibility as well as intra- and interobserver reliability. The accuracy of the absolute quantitation of MBF with factor and compartment analyses and dynamic (82)Rb PET was very good, compared with that achieved with (13)N-ammonia, for MBF of up to 2.5 mL/g/min.