Citations

ImportanceDetermining spectacle-corrected visual acuity (VA) is essential when managing many ophthalmic diseases. If artificial intelligence (AI) evaluations of macular images estimated this VA from a fundus image, AI might provide spectacle-corrected VA without technician costs, reduce visit time, or facilitate home monitoring of VA from fundus images obtained outside of the clinic.ObjectiveTo estimate spectacle-corrected VA measured on a standard eye chart among patients with diabetic macular edema (DME) in clinical practice settings using previously validated AI algorithms evaluating best-corrected VA from fundus photographs in eyes with DME.Design, Setting, and ParticipantsRetrospective cross-sectional evaluation of deidentified fundus photographs matched to spectacle-corrected VA determined by technicians on eye charts among patients with a history of DME based on optical coherence tomography and at least 2 visits within 1 to 6 months of each other at a university-based clinic between January 2014 and December 2022. Data were analyzed from January 2023 to October 2024.ExposurePreviously validated AI algorithm evaluation of fundus photographs.Main Outcomes and MeasuresAI-determined VA mean absolute error (MAE) compared with actual spectacle-corrected VA.ResultsAmong 141 patients, the mean (SD) age was 63 (13) years, 71 (50%) were male, 2 (1%) were Asian, 42 (30%) were Black or African American, and 88 (63%) were White. Among 282 eyes at visit 1, 66 had nonproliferative diabetic retinopathy (NPDR) and DME, 38 had proliferative diabetic retinopathy (PDR) and DME, 101 had NPDR and no DME, and 77 had PDR and no DME. Among 564 images (282 eyes) at both initial and follow-up visits, MAE (SD) among eyes with NPDR, with or without center-involved DME (CI-DME), was 1.16 (1.00) lines on the eye chart for VA between 20/10 and 20/20 (67 images), and 1.44 (1.15) lines for between VA 20/25 and 20/80 (231 images). MAE (SD) among eyes with PDR, with or without CI-DME, was 1.92 (1.08) lines for VA between 20/10 and 20/20 (50 images), and 1.42 (0.97) lines for spectacle-corrected VA between 20/25 and 20/80 (150 images). Only 65 images had VA 20/100 or worse, precluding meaningful analyses.Conclusions and RelevanceIn this cross-sectional study, AI evaluation of fundus photographs among patients with DME and VA 20/80 or better estimated spectacle-corrected VA within approximately 1 to 1.5 lines of actual spectacle-corrected VA. These results support use of AI evaluation of fundus photographs to determine spectacle-corrected VA among patients with DME globally, beyond ophthalmology offices.

Abstract Attempts to activate an anti-tumor immune response in glioblastoma (GBM) have been met with many challenges due to its inherently immunosuppressive tumor microenvironment. The degree and mechanisms by which molecularly and phenotypically diverse tumor-propagating glioma stem cells (GSCs) contribute to this state are poorly defined. In this study, our multifaceted approach combining bioinformatics analyses of clinical and experimental datasets, single-cell sequencing, and molecular and pharmacologic manipulation of patient-derived cells identified GSCs expressing immunosuppressive effectors mimicking regulatory T cells (Tregs). We show that this I mmunosuppressive T reg- L ike (ITL) GSC state is specific to the mesenchymal GSC subset and is associated with and driven specifically by TGF-β type II receptor (TGFBR2) in contrast to TGFBR1. Transgenic TGFBR2 expression in patient-derived GBM neurospheres promoted a mesenchymal transition and induced a 6-gene ITL signature consisting of CD274 (PD-L1), NT5E (CD73), ENTPD1 (CD39), LGALS1 (galectin-1), PDCD1LG2 (PD-L2), and TGFB1. This TGFBR2-driven ITL signature was identified in clinical GBM specimens, patient-derived GSCs and systemic mesenchymal malignancies. TGFBR2 High GSCs inhibited CD4+ and CD8+ T cell viability and their capacity to kill GBM cells, effects reversed by pharmacologic and shRNA-based TGFBR2 inhibition. Collectively, our data identify an immunosuppressive GSC state that is TGFBR2-dependent and susceptible to TGFBR2-targeted therapeutics.

Purpose: Augmented reality (AR) may allow vitreoretinal surgeons to leverage microscope-integrated digital imaging systems to analyze and highlight key retinal anatomic features in real time, possibly improving safety and precision during surgery. By employing convolutional neural networks (CNNs) for retina vessel segmentation, a retinal coordinate system can be created that allows pre-operative images of capillary non-perfusion or retinal breaks to be digitally aligned and overlayed upon the surgical field in real time. Such technology may be useful in assuring thorough laser treatment of capillary non-perfusion or in using pre-operative optical coherence tomography (OCT) to guide macular surgery when microscope-integrated OCT (MIOCT) is not available. Methods: This study is a retrospective analysis involving the development and testing of a novel image-registration algorithm for vitreoretinal surgery. Fifteen anonymized cases of pars plana vitrectomy with epiretinal membrane peeling, along with corresponding preoperative fundus photographs and optical coherence tomography (OCT) images, were retrospectively collected from the Mayo Clinic database. We developed a TPU (Tensor-Processing Unit)-accelerated CNN for semantic segmentation of retinal vessels from fundus photographs and subsequent real-time image registration in surgical video streams. An iterative patch-wise cross-correlation (IPCC) algorithm was developed for image registration, with a focus on optimizing processing speeds and maintaining high spatial accuracy. The primary outcomes measured were processing speed in frames per second (FPS) and the spatial accuracy of image registration, quantified by the Dice coefficient between registered and manually aligned images. Results: When deployed on an Edge TPU, the CNN model combined with our image-registration algorithm processed video streams at a rate of 14 FPS, which is superior to processing rates achieved on other standard hardware configurations. The IPCC algorithm efficiently aligned pre-operative and intraoperative images, showing high accuracy in comparison to manual registration. Conclusions: This study demonstrates the feasibility of using TPU-accelerated CNNs for enhanced AR in vitreoretinal surgery.

Abstract Objective Data extraction from the published literature is the most laborious step in conducting living systematic reviews (LSRs). We aim to build a generalizable, automated data extraction workflow leveraging large language models (LLMs) that mimics the real-world 2-reviewer process. Materials and Methods A dataset of 10 trials (22 publications) from a published LSR was used, focusing on 23 variables related to trial, population, and outcomes data. The dataset was split into prompt development (n = 5) and held-out test sets (n = 17). GPT-4-turbo and Claude-3-Opus were used for data extraction. Responses from the 2 LLMs were considered concordant if they were the same for a given variable. The discordant responses from each LLM were provided to the other LLM for cross-critique. Accuracy, ie, the total number of correct responses divided by the total number of responses, was computed to assess performance. Results In the prompt development set, 110 (96%) responses were concordant, achieving an accuracy of 0.99 against the gold standard. In the test set, 342 (87%) responses were concordant. The accuracy of the concordant responses was 0.94. The accuracy of the discordant responses was 0.41 for GPT-4-turbo and 0.50 for Claude-3-Opus. Of the 49 discordant responses, 25 (51%) became concordant after cross-critique, increasing accuracy to 0.76. Discussion Concordant responses by the LLMs are likely to be accurate. In instances of discordant responses, cross-critique can further increase the accuracy. Conclusion Large language models, when simulated in a collaborative, 2-reviewer workflow, can extract data with reasonable performance, enabling truly “living” systematic reviews.

Background/Objectives: CSCs are critical drivers of the tumor and stem cell phenotypes of glioblastoma (GBM) cells. Chromatin modifications play a fundamental role in driving a GBM CSC phenotype. The goal of this study is to further our understanding of how stem cell-driving events control changes in chromatin architecture that contribute to the tumor-propagating phenotype of GBM. Methods: We utilized computational analyses to identify a subset of clinically relevant genes that were predicted to be repressed in a Polycomb repressive complex 2 (PRC2)-dependent manner in GBM upon induction of stem cell-driving events. These associations were validated in patient-derived GBM neurosphere models using state-of-the-art molecular techniques to express, silence, and measure microRNA (miRNA) and gene expression changes. Advanced Poly(β-amino ester) nanoparticle formulations (PBAEs) were used to deliver miRNAs in vivo to orthotopic human GBM tumor models. Results: We show that glioma stem cell (GSC) formation and tumor propagation involve the crosstalk between multiple epigenetic mechanisms, resulting in the repression of the miRNAs that regulate PRC2 function and histone H3 lysine 27 tri-methylation (H3K27me3). We also identified miR-217-5p as an EZH2 regulator repressed in GSCs and showed that miR-217-5p reconstitution using advanced nanoparticle formulations re-activates the PRC2-repressed genes, inhibits GSC formation, impairs tumor growth, and enhances the effects of ionizing radiation in an orthotopic model of GBM. Conclusions: These findings suggest that inhibiting PRC2 function by targeting EZH2 with miR-217-5p advanced nanoparticle formulations could have a therapeutic benefit in GBM.

The Quantitative Imaging Biomarkers Alliance Profile for dynamic susceptibility contrast (DSC) MRI and its longitudinal claims for the precision of a DSC MRI–derived quantitative imaging biomarker to assess the relative cerebral blood volume changes in gliomas are summarized.

Introduction: Oncologists increasingly recognize the need to evaluate the impact of cancer therapies on patients' time, known as “treatment-related time toxicity”, which is measured as the number of days of contact with the healthcare system (Gupta A et al. J Clin Oncol 2022). Current standard-of-care options for patients with TCE RRMM require frequent clinic visits for treatment administration. Reducing treatment-related time toxicity of treatment without compromising efficacy or safety may have an impact on patients' quality of life and influence treatment choice. BCMA-directed bispecific antibodies (BCMA bsAbs) provide a novel treatment option for TCE RRMM, with response-adapted dosing frequency allowing for extended dosing intervals for patients. The frequency of extended dosing intervals varies between BCMA bsAbs according to their prescribing information or pivotal trial protocol. The aim of this analysis was to evaluate the impact of BCMA bsAbs' extended dosing intervals on treatment-related contact (TRC) days corresponding to treatment administration as a component of time toxicity. Methods: An Excel-based longitudinal model was developed to quantify the number of TRC days experienced by patients with TCE RRMM within the healthcare system when receiving treatment with BCMA bsAbs (linvoseltamab, teclistamab, or elranatamab) over 1-2 years. TRC days included all planned treatment administration days. Model inputs were sourced from product labels or pivotal trial protocols for bsAbs, as well as the published literature. To allow for a comparison of total dosing regimens, it was assumed that patients continued receiving treatment throughout the study period with no discontinuation. Dosing intervals for each product during the first 6 months were once weekly (QW), except for linvoseltamab (QW for Weeks 1-14, then once every 2 weeks [Q2W] from Weeks 16-24, per Phase 2 trial protocol). The extended dosing interval for linvoseltamab was Q4W for patients reaching very good partial response or better at or after 24 weeks, and Q2W for both teclistamab patients who maintained a complete response (CR) or better for ≥6 months and for elranatamab patients who reached partial response or better at or after 6 months. Two switching scenarios were considered for teclistamab: an earliest possible switch (at Week 26), and a time-varying switch based on digitized time to CR data from MajesTEC-1 (Usmani SZ et al. ASCO 2023). Results were calculated to evaluate the difference in TRC days between the scenarios with and without extended dosing intervals for each treatment. Results: In the first year of treatment, the total number of TRC days with and without extended dosing intervals was lowest with linvoseltamab. TRC days with linvoseltamab were reduced from 33 to 26 days with extended dosing intervals. Next lowest was elranatamab (53 to 39 days), followed by teclistamab (54 to 41 days for earliest possible switch and 54 to 50 days with time-varying switching). Over 2 years, linvoseltamab had the lowest number of TRC days (59 to 39 days), followed by elranatamab (105 to 65 days) and teclistamab (106 to 67 days for earliest possible switch and 106 to 80 days with time-varying switching). Conclusions: Extended dosing intervals with BCMA bsAbs can substantially reduce time toxicity for patients with RRMM. Linvoseltamab had the lowest number of TRC days among BCMA bsAbs.

Introduction: Mantle cell lymphoma (MCL) is an uncommon B-cell non-Hodgkin lymphoma (NHL) Prognosis is highly variable due to underlying biological and molecular heterogeneity. The first-line (1L) treatment paradigm for MCL continues to evolve with increasing use of bruton tyrosine kinase inhibitors (BTKi) and decreasing use of autologous stem cell transplant (ASCT). TP53 alterations are associated with inferior outcomes in MCL. The aim of this retrospective study is to determine treatment strategies and clinical outcomes in MCL patients based on TP53 gene mutation status. Methods: Retrospective data were collected from 19 institutions in the United States from 1/1993 to 1/2024. We examined treatment patterns and outcomes among MCL patients >18 years of age with known TP53 mutation status by sequencing. Patients who did not receive treatment were excluded. The analysis was stratified based on presence of TP53 mutations. Systemic therapies were categorized as chemo-immunotherapy (CIT), non-chemotherapy (including monoclonal antibodies, lenalidomide, venetoclax, and novel agents on clinical trial), and regimens containing BTKi. The primary outcome was overall survival (OS) and secondary outcomes were progression free survival (PFS) and time to next treatment (TTNT). OS and PFS were estimated with the Kaplan-Meier method and log-rank tests were used to compare groups. Results: 645 patients were included in the study, with 468 men (72.6%) and median age was 64 years (range 30-89). 230/645 patients (35.7%) had a TP53 mutation, of which 182 (79%) were identified prior to 1L. While the male/female distribution was the same between the mutated and unmutated groups, median age at diagnosis and MIPI risk score was higher in the TP53 mutated group (age 65.9 vs 63.1 yrs, p= 0.015; MIPI 6.4 vs 6.1, p<0.001). TP53 unmutated patients had a median of 1 line of therapy vs. 2 for the mutated group (p<0.001, range 1-8). Blastoid histology and higher ki67 were also associated with the TP53 mutated group (p=0.045, 0.01 respectively) The median OS was 14.2 years for TP53 unmutated and 8.3 years for TP53 mutated patients (p<0.0001). PFS in 1L was 0.95 years in TP53 mutated compared to 2.02 years in TP53 unmutated group (p<0.0001). Selection of 1L therapeutic strategy was not associated with presence of TP53 mutation (Χ2 p= 0.21) although more patients in the TP53 unmutated group underwent ASCT (119/415 vs 38/230; Χ2 p=0.0006) TP53 mutation was associated with inferior survival compared to TP53 unmutated patients for CIT in 1L but the difference did not reach statistical significance when non-chemotherapy regimens and regimens containing BTKi were used. For CIT in 1L, 5 year OS was 78% [71-84%] for unmutated compared to 58% [48-67%] for the mutated group (p=0.0001 ). For non-chemotherapy regimens, 5 year OS for unmutated patients was 73% [52- 86%] compared to 52% [25-73%] for the mutated group (p=0.09). For regimens containing BTKi, 5 year OS for unmutated was 93% [81-97%] compared to 57% [28-78%] for mutated (p=0.07). Within the TP53 mutated group, there were no differences in any clinical outcomes between those receiving CIT, non-chemotherapy, or BTKi in 1L. While median OS appeared numerically longer in the BTKi group (14.4 yrs vs 8.8 yrs for CIT and 6.8 yrs for non-chemotherapy), the difference did not reach statistical significance (p=0.58). In contrast,TP53 unmutated patients who received CIT in 1L had better TTNT (p=0.0039) compared to those in non-chemotherapy and BTKi groups. There was no improvement in PFS (p=0.79) or OS (p=0.47). Conclusions To our knowledge this is the largest retrospective real world dataset of MCL treatment outcomes stratified by TP53 mutation. As in prior studies, TP53 mutation portends a worse prognosis in MCL compared to TP53 unmutated patients, but outcomes are better than historically reported. CIT in 1L was not associated with improved OS in either group. Irrespective of TP53 mutation status, incorporation of BTKi in first line was not associated with improved clinical outcomes. The gap in OS between TP53 mutated and unmutated MCL may be narrowing as treatment paradigms shift away from intensive induction chemotherapy and ASCT, and due to an increasing number of effective therapies for relapsed patients. Clinical trials of novel agents are needed to further improve the OS of TP53 mutated MCL patients.

Multiple myeloma (MM), an incurable hematologic malignancy, presents significant challenges in managing its recurrence. The bone marrow tumor microenvironment (BM-TME) is critical to MM progression. However, classical immune checkpoint blockade (ICB) therapies, like anti-PD1, have shown limited efficacy in MM. Therefore, elucidating the compositional features and regulatory mechanisms of the TIME in MM is crucial for developing novel immunotherapeutic targets. Tissue-resident memory CD8+ T cells (CD8+ TRMs) reside indefinitely at multiple anatomical locations like the skin, intestine, and bone marrow, functioning as frontline responders against reencountered pathogens and tumors. Whether CD8+ TRMs within BM-TME are involved in the MM cell immune escape and pathogenesis has not been systematically elucidated. Firstly, we leverage single-cell RNA sequencing (scRNA-seq) to dissect the BM CD8+ T cells across patients at different stages of MM, and identify that CD69+CD8+ T subsets were BM residents, and the proportion of BM CD8+ TRMs between health donors and MM patients was comparable. Moreover, our findings indicated that BM CD8+ TRMs exhibit exhausted features and enrichment of NK genes with MM progresses. Importantly, we observed a considerable elevation of the NK gene KLRB1 and its encoded protein CD161 in BM CD8+ TRMs in MM patients. Immunofluorescence labeling of BM biopsy samples from MM patients confirmed the infiltration of the BM-TME with CD3+CD8+CD69+CD161+ cells. Then, we found that KLRB1 expression in BM CD8+ TRMs was higher than that of other classical inhibitory receptor genes (e.g., PDCD1, CTLA4, LAG3), and CLEC2D (encodes the ligand of CD161) was also markedly expressed in patients' MM cells, exhibiting considerably higher than classical immunosuppressive receptor ligand genes (e.g., PDL1, CD80, CD86). Moreover, patients with higher CLEC2D expression had unfavorable overall survival. We further found that the CD161 expression was higher in CD69+CD8+ TRMs compared to circulating CD69-CD8+ T cells, and the percentage of BM CD161+CD8+ TRMs in MM patients was significantly higher than in healthy donors, while there was no difference for PD-1, LAG3, and TIM3 expression. Notably, only a small fraction of BM CD8+ TRMs have mucosal-associated invariant T cell phenotypes (identified by CD161+TCR-Vα7.2+). Additionally, the CLEC2D expression of BM cells was analyzed, and the highest level of CLEC2D expression was observed in B cells from MM patients, followed by plasma cells. Importantly, CLEC2D expression in MM patients was significantly higher than in plasma cells from healthy donors. Next, to investigate whether the blockade of CD161 could restore the function of CD8+ TRMs, the primary BM cells from MM patients were isolated and stimulated ex vivo. Compared to the anti-PD-1 and isotype group, CD161 blocking presents enhanced CD8+ TRMs proliferation and cytokine secretion levels (TNF-α, IFN-γ, IL2), increasing MM cells apoptosis. Chimeric antigen receptor T (CAR-T) cell therapy has achieved impressive success in treating relapsed/refractory MM. How to prevent CAR-T cell exhaustion and enhance function remains a significant challenge. Therefore, we next set out to investigate the impact of CD161 on CAR-T cells. Our results demonstrated that CD161 was highly expressed in early peripheral blood (PB) CAR-T cells after infusion in MM patients but decreased with time. Moreover, the CD161 expression of CAR-T cells was considerably higher than endogenous T cells. Notably, BM-derived CAR-T cells have higher CD161 expression than PB counterparts for the same patient, suggesting CD161 may have BM-specific expression enrichment. After co-culturing MM and CAR-T cells in vitro for 24 hours, we found that blocking CD161 led to an augmented proportion of CAR-T cells, increased IFN-γ secretion, and a notable elevation in tumor cell apoptosis compared to the anti-PD-1 and isotype groups. Taken together, we demonstrated that CD8+ TRMs constitute a vital anti-tumor cell subset in the MM BM-TME, which exhibits a high expression of the inhibitory CD161 receptor, binding to the CLEC2D expressed by MM cells, thereby mediating the dysfunction of CD8+ TRMs. Blocking the CD161-CLEC2D pathway is expected to become a novel target for MM immunotherapy. The combination of CD161 inhibitors and CAR-T cell therapy can potentially enhance the anti-myeloma function of CAR-T cells.

Background: Polatuzumab vedotin (Pola) is an antibody drug conjugate targeting CD79b that incorporates the microtubule disrupting monomethyl auristatin E payload. Pola was approved in 2023 for the upfront treatment of diffuse large B cell lymphoma (DLBCL) in combination with rituximab, cyclophosphamide, doxorubicin, and prednisone (polaRCHP) based on the POLARIX trial. However, the activity of pola appears to vary based on DLBCL cell of origin (COO) subtype, with greater efficacy in activated B cell (ABC) DLBCL compared to germinal center B cell (GCB) DLBCL (Russler-Germain Blood 2023). When outcomes in POLARIX are stratified by COO, polaRCHP demonstrated progression free survival (PFS) and overall survival (OS) benefits over RCHOP in patients with ABC but not GCB subtype DLBCL (PFS HR 0.34 ABC vs 1.03 GCB, OS HR 0.27 ABC vs 1.64 GCB; Morschhauser ASH 2023 and FDA data). Most prospective studies of pola used gene expression to assess COO, which is not readily clinically available. This study aims to examine if COO subtype by Hans algorithm (based on CD10, BCL6, and MUM1 expression by immunohistochemistry) is predictive of pola responses and outcomes in patients with large B cell lymphoma (LBCL). Methods: We conducted a retrospective, multicenter observational study of adults with LBCL who received pola (monotherapy or in combination) from 2015-2024 with Hans algorithm COO data and response assessment available. Patients were stratified by receipt of pola in the frontline vs relapsed/refractory (R/R) setting. The primary endpoint was overall response rate (ORR) to pola-based treatment in GCB vs non-GCB subtypes. Secondary endpoints included complete response rate (CRR), PFS, and OS. For time to event endpoints, patients responding to pola who received CAR T or stem cell transplantation prior to progression were censored at time of cellular therapy. Demographics were summarized by descriptive statistics. Comparative and survival analyses used Chi squared/Fischer exact tests and Kaplan Meier analysis. Data cutoff was July 1, 2024. Results: 718 patients were included. Median follow-up was 15 months (interquartile range [IQR] 8-32). Median age at LBCL diagnosis was 66 years (IQR 56-74); 62% were men; 69 (10%) had double hit lymphoma (DHL); 129 (18%) had underlying low grade lymphoma. Pola was used in the frontline setting in 293 (41%) patients, of whom 139 (47%) had GCB vs 154 (53%) non-GCB COO. Of the 426 patients receiving pola for R/R LBCL (including one patient receiving pola retreatment), 224 (53%) had GCB vs 202 (47%) non-GCB COO. Among patients receiving pola-based regimens in the R/R setting (GCB vs non-GCB: pola monotherapy = 7% vs 4%, pola + R = 26% vs 26%, pola + BR = 55% vs 61%, pola + other = 12% vs 9%; p = 0.44), the ORR in GCB vs non-GCB LBCL was 36% vs 61% (p < 0.001), while CRR in GCB vs non-GCB was 17% vs 36% (p < 0.001). COO remained a significant component of logistic multivariate regression models of ORR (p < 0.001) and CRR (p < 0.001) when incorporating IPI at original LBCL diagnosis, histology, DHL status, and refractory status to frontline therapy. Despite this, no significant PFS or OS differences by COO were seen, potentially due to approximately 1/4 patients being censored due to receiving subsequent cellular therapy (e.g. CAR T) prior to progression. In patients receiving frontline pola-based regimens (of which 275 [94%] included an anthracycline), ORR and CRR were similar in GCB vs non-GCB (ORR 89% vs 88% p > 0.5; CRR 76% vs 77% p > 0.5). While PFS and OS were similar between COO groups in the frontline cohort, follow-up and cohort size were limited. Conclusion: Based on this real-world evidence, COO classification by Hans algorithm is a strong predictor of the activity of pola-containing therapy in R/R LBCL, with significantly higher ORR and CRR in patients with non-GCB compared to GCB COO LBCL. This aligns with the cumulative existing evidence. ORR and CRR to pola-containing therapy in the frontline setting did not differ significantly by COO subtype, but longer follow-up of larger cohorts is needed to clarify the predictive role of COO by Hans algorithm in this setting, especially considering the PFS advantage with polaRCHP in POLARIX was not associated with significant differences in ORR or CRR. Overall, clinicians should consider factoring COO subtype into treatment decisions regarding pola-containing regimens, and studies of pola-based therapies should stratify patients and report outcomes by COO subtype.