Citations

Abstract Purpose To develop and implement an in‑house treatment‑management platform to improve communication, data accessibility, and task tracking across a radiation oncology department. Methods We developed Capsule, a custom software platform that centralizes patient data from the ARIA oncology information system (Varian Medical Systems, Palo Alto, United States), and integrates it with data from Outlook (Microsoft, Redmond, United States), spreadsheets, and manual inputs. Built using C#, SQL, and the .NET framework, it features a graphical interface in XAML with twelve modular tools to support workflows across a large multi‐modality department. A formal validation, feedback, and maintenance cycle was established. Its impact on workflow efficiency was evaluated using average CarePath task duration before and after implementation, supplemented by user interviews and surveys. Results The Department Schedule, Treatment Planning and On‐Treatment Whiteboard, Spreadsheet, and various workload tracking modules emerged as the most frequently used among dosimetrists surveyed.The Whiteboard module consolidates planning metadata for streamlined workflows. Workload and scheduling modules help distribute tasks fairly. The Department Schedule module centralizes daily role and contact information. The Spreadsheet module allows dosimetrists to identify a reference for unique plans and provides secure access to large exportable tabulated patient data and clinical trends. Other modules support proton down‐time replanning and special procedure physics assignment. One year after Capsule's implementation, most tasks in the external beam radiation therapy workflow experienced a reduction in average duration. Discussion Capsule streamlines workflows by integrating multiple tools into a single platform, supporting urgent scenarios, and improving task hand‐off. Its modular design allows customization to local workflows, but implementation must be guided by institutional oversight and formal quality management to ensure data security and compliance. Conclusion Capsule can provide a solution for improved workflow efficiency, communication, and interdisciplinary coordination. By democratizing access to routine and retrospective data, it supports treatment planning, scheduling, and quality improvement. Ongoing updates are essential as clinical practices evolve.

Abstract Iron-deficiency anemia (IDA) affects nearly one-third of women globally, with heavy menstrual bleeding (HMB) being a significant risk factor among those of reproductive age. Despite the substantial burden of HMB-related IDA, patients often face delays in receiving effective treatment. Oral iron is typically used as first-line therapy, despite frequent gastrointestinal side effects and challenges in treatment adherence. This study evaluated the reproductive life span cost-effectiveness of first-line intravenous (IV) vs oral iron therapy for women with HMB and IDA. We developed a Markov model simulating a cohort of women aged 18 to 51 years, comparing first-line treatment with IV iron dextran, IV ferumoxytol, IV iron sucrose, or oral ferrous sulfate. Costs and quality-adjusted life years (QALYs) were estimated from the societal perspective. IV iron dextran was the cost-effective treatment, yielding 19.26 QALYs at a cost of $157 500, compared with 19.10 QALYs at $152 900 for oral ferrous sulfate. The incremental cost-effectiveness ratio for IV iron dextran was $28 600 per QALY. Treatment with IV ferumoxytol and iron sucrose cost $158 300 and $163 500, respectively, and did not provide additional QALY benefit compared with IV iron dextran, which remained the cost-effective treatment across a range of scenarios and sensitivity analyses. Our findings indicate that first-line treatment with IV iron dextran is the cost-effective strategy for managing IDA in women with HMB at commonly accepted willingness-to-pay thresholds. These results support expanding access to IV iron as a first-line option, and highlight the need to reduce treatment delays and insurance-related barriers.

Abstract Background Apolipoprotein E (apoE) facilitates the fibrillization and deposition of amyloid β (Aβ) in Alzheimer's Disease (AD). Supported by robust literature from various groups, in our current study we have investigated the relative amyloidogenic properties of the three major human APOE alleles. Knowledge from such studies could provide insight into the role of ApoE protein in the deposition, evolution and maturation of parenchymal Aβ. Method We used hAbetaSAA (Jax Labs) mice crossed with human APOE targeted replacement mice to create hAbetaSAA mice homozygous for human APOE E2, E3 and E4 . Furthermore, APPsi mice (hAPPswe/indiana: Borchelt) expressing human APOE E3 or E4 were injected intracerebrally with AD brain lysates to examine the role of APOE genotype on Aβ seeding. Neuropathological examination was conducted via immunohistochemistry, immunofluorescence and immunoblotting. Result hAbetaSAA mice expressing human APOE E4 showed higher Aβ deposition relative to APOE E3 and APOE E2 by 5 months of age and progressing through 8 months and 13 months of age. In comparison, APP mice expressing humanized Aβ (Jax Labs) with human APOE E4 did not develop deposits in their lifetime. hAbetaSAA mice with endogenous mouse Apoe had the highest Aβ plaque deposition, with an approximately 60‐fold increase in Aβ burden compared to SAA APP mice expressing APOE E3 , especially in the earlier deposition phase. Removing mouse Apoe resulted in dose‐dependent reduction in Aβ deposition, consistent with previously published literature. Plaque morphology did not show any notable differences between APOE genotypes, and colocalization of apoE with Aβ were relatively equivalent across different APOE genotypes. Further, in the APPsi model seeded with AD‐derived seeds, we found that mice with APOE E4 or APOE E3 developed equal burdens of diffuse parenchymal Aβ, with no detectable changes in plaque morphology. Conclusion We conclude that apoE variants broadly influence the rate of amyloid deposition, but do not modify the morphology of the resulting deposits or colocalization with Aβ. Notably, when seeding with AD patient brain lysate, there was no significant difference in Aβ burden between APPsi mice expressing APOE E3 and E4 . Our study adds insights into the role of apoE isoforms in determining Aβ‐related phenotype.

To improve synthetic CT (sCT) generation from cone-beam CT (CBCT) in radiotherapy, we propose a multiscale segmentation-guided diffusion framework. The proposed model integrates anatomical priors across multiple spatial resolutions through a segmentation mask pyramid and introduces a scale-specific loss function to guide learning at each level. When evaluated on the SynthRAD2023 brain dataset, our model achieves a mean absolute error (MAE) of 61.82 HU, a peak signal-to-noise ratio (PSNR) of 32.05 dB, and a structural similarity index (SSIM) of 0.90, outperforming baseline models. These results suggest that multiscale anatomical guidance can improve the fidelity and anatomical consistency of sCT images, thus facilitating high-quality CBCT-to-CT translation in radiotherapy applications.

Abstract Introduction: Monoclonal gammopathy of undetermined significance (MGUS) affects over 2 million individuals aged ≥70 in the US. Earlier detection of smoldering multiple myeloma (SMM) and multiple myeloma (MM) in individuals with suspected MGUS via bone marrow biopsy allows for earlier active monitoring and potential intervention, including with daratumumab for high-risk SMM. Risk prediction for the presence of SMM or MM with the iStopMM and Mayo Clinic models leads to decreased individual-patient and population-level resource utilization burden of bone marrow testing with the tradeoff of some missed non-MGUS diagnoses. We sought to quantify the value of earlier diagnosis, while additionally accounting for the recent results of the AQUILA trial, by assessing the lifetime cost-effectiveness of bone marrow testing strategies for patients with suspected MGUS. Methods: In this independent analysis free of industry influence, we used a decision tree with nested Markov cohort models to evaluate diagnosis and disease progression from MGUS and SMM to MM across 3 diagnostic strategies employing: (1) iStopMM risk prediction model versus (2) Mayo Clinic risk stratification and prediction model versus (3) a test-all strategy. Both risk prediction model performance characteristics were derived from the largest population-level prospective cohort study. Individuals in the MGUS state were monitored annually for disease progression with blood tests. The SMM state included two groups: those correctly diagnosed through biopsy and those misclassified as MGUS due to not undergoing biopsy under the prediction model triage rules (missed SMM diagnosis). As recommended by guidelines, diagnosed patients with non–low-risk SMM were monitored with blood tests every three months; those with low-risk SMM were tested every six months. All diagnosed SMM patients underwent imaging for bone lesions at diagnosis and annually for up to five years. Missed SMM patients were monitored at the same frequency as those with MGUS until disease progression. Early versus delayed myeloma diagnosis complications were informed by real-world multi-institutional quasi-experimental data of consecutively diagnosed patients with MM. MM treatment and outcomes across up to four lines of therapy were modeled exactly as previously published in a validated model. In additional scenario analyses, we assessed the impact of daratumumab monotherapy in the care of people with high-risk SMM across all 3 testing strategies by incorporating the progression free survival and overall survival benefit as reported in the phase 3 AQUILA study. The primary outcome was the incremental net monetary benefit (iNMB), reported across all accepted US willingness-to-pay (WTP) thresholds from a US health system perspective. We concluded with extensive deterministic sensitivity analyses and additionally quantified uncertainty across all model parameters simultaneously with 10,000 Monte Carlo simulations. Results: The iStopMM model was the cost-effective strategy yielding the highest NMB of all strategies at every WTP threshold, with per-patient cost and effectiveness of $188,300 and 10.21 QALYs, respectively. This strategy was dominant—costing less and producing more QALYs—than both the Mayo Clinic and test-all strategies. At the population level, use of the iStopMM model over the Mayo Clinic model and the test-all strategy yields iNMBs of $0.67 [95% CI: 0.2 – 1.14] billion and $3.13 [95%CI: 2.80-3.47] billion, respectively, in favor of the iStopMM strategy. The iStopMM strategy was the cost-effective strategy in 100% of 10,000 second order Monte Carlo iterations. In deterministic sensitivity analyses, the top three parameters the iNMB was most sensitive to were prediction model test characteristics (i.e., sensitivity and specificity) as well as the baseline prevalence of SMM. However, no parameter variation could change the conclusion that iStopMM is the cost-effective strategy. In scenario analyses incorporating benefit of daratumumab monotherapy in the care of individuals with high-risk SMM, the iStopMM model remained the cost-effective strategy. Conclusion: Using the iStopMM risk prediction model for marrow sampling in patients with suspected MGUS leads to cost savings per patient evaluated, both with and without incorporating daratumumab benefit in the care of individuals with high-risk SMM, as compared to both the Mayo Clinic model and testing all with bone marrow biopsy.

With the rapid development of artificial intelligence (AI), various deep learning (DL) methods have been introduced into radiation oncology. Among them, the generation of synthetic Computed Tomography (sCT) images has attracted increasing attention, as it supports different clinical scenarios, from image-guided adaptive radiotherapy (IGART) to the simulation-free workflow. This review provides a comprehensive overview of recent studies on DL-based sCT synthesis in radiotherapy from multiple imaging modalities, including Cone-Beam CT (CBCT), Magnetic Resonance Imaging (MRI), and diagnostic CT, and discusses their clinical applications in CBCT-based online adaptive radiotherapy, MRI-guided radiotherapy, and simulation-free workflows. We also examine the architectures of representative DL models such as convolutional neural networks (CNNs) and generative adversarial networks (GANs) and summarize emerging training strategies. Finally, we discuss current challenges of clinical translation of DL algorithms into clinical practice and suggest potential directions for future research. Overall, this paper highlights the potential of AI-driven sCT generation to advance treatment planning by reducing imaging burden, improving dose accuracy, and accelerating workflow efficiency, thus ultimately improving the treatment outcome of patient care.