Citations

Abstract Background Diabetic retinopathy (DR) is the leading cause of preventable blindness among working-age adults worldwide, yet screening coverage remains inadequate, particularly in low-and middle-income countries. Automated deep learning systems offer potential to address the global shortage of expert graders, but most existing models lack lesion-level interpretability and are not aligned with established clinical referral frameworks. We developed and validated DRAGS (Diabetic Retinopathy Automated Grading System), a hybrid deep learning model that grades DR according to the UK Diabetic Eye Screening Programme (DESP) classification and provides lesion-level explainability. Methods We trained and validated a DenseNet-201-based convolutional neural network on 20,281 anonymised fundus images from two tertiary eye care institutions in Bangladesh. Images were graded by fellowship-trained retinal specialists using the UK DESP framework, resulting in 10 clinically interpretable classes that combine retinopathy grade (R0–R3) and maculopathy status (M0/M1). A companion dataset of 2,936 pixel-level lesion masks spanning nine pathological categories was used to train a parallel multi-label lesion-detection head. The dataset was partitioned 70:15:15 (patient-stratified). Performance was evaluated using macro-averaged AUROC (DeLong estimator), sensitivity, specificity, F1 score, quadratically weighted Cohen’s κ, and expected calibration error (ECE), with 95% CIs from 2000 bootstrap resamples. Grad-CAM spatial alignment with ground-truth lesion masks was assessed using Dice and IoU. This study follows the TRIPOD+AI reporting guidelines. Findings On the held-out test set (Component I: n = 3,044; Component II: n ≈ 440), DRAGS achieved class-wise precision, recall, and F1 scores ranging from 0·88 to 0·99 across all ten UK DESP grades, with advanced proliferative stages (R3–M0, R3–M1) consistently exceeding 0·95. Overall accuracy was approximately 91·1% and quadratically weighted Cohen’s κ was approximately 0·90. For referable versus non-referable DR, sensitivity was 90·7% and specificity was 91·9%. The companion lesion-detection head achieved macro-averaged sensitivity of 93·9%, specificity of 99·5%, and AUC of 0·997 across nine lesion classes; seven of nine classes achieved AUC = 1·00. Grad-CAM activations showed progressive spatial shift from diffuse (normal) to lesion-dense peripheral patterns (proliferative DR), with maximal agreement for microaneurysms and exudates. Mean inference time was 110–160 ms per image. Interpretation DRAGS demonstrates high diagnostic accuracy for nine-class UK DESP-aligned DR grading, with clinically interpretable lesion-level explainability on a large real-world LMIC dataset. External validation and prospective clinical evaluation are warranted before deployment. Funding The present study received no funding.

Abstract Clonal hematopoiesis (CH) is characterized by expanding blood cell clones carrying somatic mutations in healthy aged individuals and is associated with various age-related diseases and all-cause mortality. While CH mutations affect diverse genes associated with myeloid malignancies, their mechanisms of expansion and disease associations remain poorly understood. We investigate the relationship between clonal fitness and clinical outcomes by integrating data from three longitudinal aging cohorts (n=713, observations=2,341). We demonstrate pathway-specific fitness advantage and clonal composition influence clonal dynamics. Further, the timing of mutation acquisition is necessary to determine the extent of clonal expansion reached during the host individual’s lifetime. We introduce MACS120, a metric combining mutation context, timing, and variant fitness to predict future clonal growth, outperforming traditional variant allele frequency measurements in predicting clinical outcomes. Our unified analytical framework enables standardized clonal dynamics inference across cohorts, advancing our ability to predict and potentially intervene in CH-related pathologies.

Introduction ABBV-383 is a differentiated BCMA x CD3 bispecific antibody (fully humanized IgG4) T-cell engager composed of a bivalent BCMA binding domain with high avidity, a low-affinity CD3-binding domain, and a present but silenced-Fc tail retaining the FcRn binding allowing for an extended half-life and thereby, enabling dosing convenience. ABBV-383 monotherapy has shown promising activity and improved CRS safety profile with modified dexamethasone (Dex) premedication in patients (pts) with RRMM (NCT03933735; Rodriguez et al., JCO 2024;42[suppl 16]:7531). Here, ER and CRS PK/PD analyses supporting the selection of optimal SUD for ABBV-383 are reported. Methods ER-efficacy (N=286) and -safety (N=290) analyses were conducted using logistic regression methodology (R version 4.3.2) combining data from escalation (0.025-120 mg Q3W) and expansion (20, 40, 60 mg Q3W and 60 mg Q4W) cohorts of phase 1 study NCT03933735 and Arm A of phase 1b study NCT05650632 evaluating step-up dose (2 or 4 mg on Day 1) with a full dose (60 mg on Day 4) in pts with RRMM. Pts received modified Dex with full dose in expansion cohorts of 60 mg Q4W and 2/60 mg Q4W. In all other cohorts with or without SUD, pts received low Dex. Based on prior knowledge (JCO 2024;42[suppl 16]:7541), the safety endpoints (CRS: any-grade and ≥ G2) were modeled with Cycle 1 total (unbound + bound to soluble BCMA) Cmax, while objective response rate (ORR) was modeled with free (unbound + partially bound to soluble BCMA) Cavg. CRS ER models included a priori selected Dex (low vs. modified) and SUD (yes vs. no) as predictors. Relevant pt-specific covariates were tested in all models. CRS PK/PD analyses included total PK, IL-6, and CRS data (N=199) from the first 3 cycles of the above phase 1/1b studies. A semi-mechanistic indirect response PK/IL-6 model was developed to correlate total PK and IL-6 (NONMEM 7.5). Maximal IL-6 concentrations were derived using post hoc estimates from the PK/IL-6 model and correlated with CRS events (logistic regression; R version 4.3.2). The models included effects of time-dependent Dex (low vs. modified) and SUD (yes vs. no). The overall modeling approach was assessed by comparing observed and predicted CRS rates. Simulations were conducted for various dosing regimens, including one or two SUD and Dex premedication scenarios to support the selection of the optimal SUD. Results Consistent with clinical observations (separate abstract submission), any-grade CRS model predicted probabilities suggested two key outcomes: 1) premedication with modified Dex improves the CRS safety profile compared to the low Dex (42% vs 69% for 60 mg Q4W); 2) modified Dex with SUD further improves the CRS safety profile of ABBV-383 compared to low Dex with SUD (29% vs 43% for 2/60 mg Q4W). Consistent with prior work, no ER relationship was observed for ≥ G2 CRS (p > 0.05). The ORR model resulted in a predicted probability of 68% ORR (closely matching with overall observed 64% ORR) with no impact of SUD and modified Dex. Only soluble BCMA was selected as a covariate in the ORR ER model. The final semi-mechanistic PK/IL-6 model well captured the observed IL-6 data, and the CRS rates (predicted by the logistic regression models) closely matched with the observed data. The CRS PK/PD analyses findings were consistent with the clinical observations and ER analyses key outcomes 1 and 2. The CRS PK/PD analyses predictions comparing the 1-SUD vs 2-SUD (untested) regimens with modified Dex suggested that 1) among the 2-SUD regimens with modified Dex, the 2/3/60 mg regimen would have the maximum reduction of 4% points in any-grade CRS compared to the 2/60 mg 1-SUD regimen, however, with substantial overlap of confidence intervals and with no decrease in ≥ G2 CRS (< 0.5% point); 2) 4 mg as an intermediate SUD in a 2-SUD regimen would lead to higher any-grade and ≥ G2 CRS rates (~9% and 3% points higher, respectively), compared to the 2/60 mg Q4W regimen. Conclusions ER and CRS PK/PD analyses suggested that modified Dex and SUD improve the CRS safety profile of ABBV-383 with no impact on efficacy (ORR) in pts with RRMM. The CRS PK/PD predictions indicated that implementation of 2-SUDs provide minimal additional or no improvement in CRS safety profile compared to 1-SUD. Overall, the 1-SUD dosing regimen of 2 mg with modified premedication prior to the full dose of 60 mg Q4W showed promising results and can be considered for future evaluation and studies.

Abstract Background Abuse of fentanyl and its analogs is a major contributor to the opioid overdose epidemic in the United States, but detecting and quantifying trace amounts of such drugs remains a challenge without resorting to sophisticated mass spectrometry-based methods. Methods A sensitive immunoassay with a sub-picogram limit of detection for fentanyl and a wide range of fentanyl analogs has been developed, using a novel high-affinity antibody fused with NanoLuc, a small-size luciferase that can emit strong and stable luminescence. When used with human urine samples, the assay has a sub-picogram limit of detection for fentanyl, with results fully concordant with LC-MS. Results When applied to clinical samples, the novel chemiluminescence immunoassay can detect low positive fentanyl missed by routine screening immunoassays, with a limit of detection of 0.8 pg/mL in human urine. When applied to environmental samples, the assay can detect levels as low as 0.25 pg fentanyl per inch2 of environment surface. Assay turnaround time is less than 1 h, with inexpensive equipment and the potential for high-throughput automation or in-field screening. Conclusions We have established a novel assay that may have broad applications in clinical, environmental, occupational, and forensic scenarios for detection of trace amounts of fentanyl and its analogs.

ImportanceClonal hematopoiesis of indeterminate potential (CHIP) may contribute to the risk of atrial fibrillation (AF) through its association with inflammation and cardiac remodeling.ObjectiveTo determine whether CHIP was associated with AF, inflammatory and cardiac biomarkers, and cardiac structural changes.Design, Setting, and ParticipantsThis was a population-based, prospective cohort study in participants of the Atherosclerosis Risk in Communities (ARIC) study and UK Biobank (UKB) cohort. Samples were collected and echocardiography was performed from 2011 to 2013 in the ARIC cohort, and samples were collected from 2006 to 2010 in the UKB cohort. Included in this study were adults without hematologic malignancies, mitral valve stenosis, or previous mitral valve procedure from both the ARIC and UKB cohorts; additionally, participants without hypertrophic cardiomyopathy and congenital heart disease from the UKB cohort were also included. Data analysis was completed in 2023.ExposuresCHIP (variant allele frequency [VAF] ≥2%), common gene-specific CHIP subtypes (DNMT3A, TET2, ASXL1), large CHIP (VAF ≥10%), inflammatory and cardiac biomarkers (high-sensitivity C-reactive protein, interleukin 6 [IL-6], IL-18, high-sensitivity troponin T [hs-TnT] and hs-TnI, N-terminal pro–B-type natriuretic peptide), and echocardiographic indices.Main Outcome MeasureIncident AF.ResultsA total of 199 982 adults were included in this study. In ARIC participants (4131 [2.1%]; mean [SD] age, 76 [5] years; 2449 female [59%]; 1682 male [41%]; 935 Black [23%] and 3196 White [77%]), 1019 had any CHIP (24.7%), and 478 had large CHIP (11.6%). In UKB participants (195 851 [97.9%]; mean [SD] age, 56 [8] years; 108 370 female [55%]; 87 481 male [45%]; 3154 Black [2%], 183 747 White [94%], and 7971 other race [4%]), 11 328 had any CHIP (5.8%), and 5189 had large CHIP (2.6%). ARIC participants were followed up for a median (IQR) period of 7.0 (5.3-7.7) years, and UKB participants were followed up for a median (IQR) period of 12.2 (11.3-13.0) years. Meta-analyzed hazard ratios for AF were 1.12 (95% CI, 1.01-1.25; P = .04) for participants with vs without large CHIP, 1.29 (95% CI, 1.05-1.59; P = .02) for those with vs without large TET2 CHIP (seen in 1340 of 197 209 [0.67%]), and 1.45 (95% CI, 1.02-2.07; P = .04) for those with vs without large ASXL1 CHIP (seen in 314 of 197 209 [0.16%]). Large TET2 CHIP was associated with higher IL-6 levels. Additionally, large ASXL1 was associated with higher hs-TnT level and increased left ventricular mass index.Conclusions and RelevanceLarge TET2 and ASXL1, but not DNMT3A, CHIP was associated with higher IL-6 level, indices of cardiac remodeling, and increased risk for AF. Future research is needed to elaborate on the mechanisms driving the associations and to investigate potential interventions to reduce the risk.

Introduction: Aging and inflammation are associated with risk of atrial fibrillation (AF), to which clonal hematopoiesis of indeterminate potential (CHIP) may contribute. We investigated whether CHIP was associated with AF, inflammatory and cardiac biomarkers and cardiac structural changes. Methods: We ascertained CHIP (variant allele frequency, VAF ≥2%) in the Atherosclerosis Risk in Community Study (ARIC visit-5, n = 4,131) and the UK Biobank (UKBB, n = 195,851) participants. Using multivariable Cox regression models, we studied the association of CHIP and the top three implicated genes (i.e., DNMT3A , TET2 , ASXL1 ) with incident AF and conducted a meta-analysis on both cohorts. We also ascertained high-sensitivity C-reactive protein (hs-CRP), interleukin 6 (IL6), IL18, high-sensitivity troponin T and I (hs-TnT and hs-TnI), NT-proBNP, and echocardiography indices in ARIC. We first standardized the natural log of all markers (st-ln) and used the adjusted quantile regression to compare the median in those with and without CHIP. Results: In ARIC, the mean ± SD age was 75.8 ± 5.2 years, 41% were men, 23% were Black, 25% had CHIP, and 11.6% had large CHIP (VAF ≥10%). These measures in UKBB were 56.4 ± 8.1 years, 45%, 2%, 5.8%, and 2.6%, respectively. In the meta-analysis, compared to those without CHIP, the hazard ratios (95% confidence interval) for AF were 1.12 (1.01 - 1.25, P =0.035), 1.31 (1.07 - 1.60, P =0.009), and 1.48 (1.05 - 2.08, P =0.026) for Large-CHIP, Large- TET2, and Large- ASXL1 , respectively. hs-CRP and IL18 levels were similar in those with and without CHIP as well as gene-specific subgroups. Large- TET2 and ASXL1 were associated with higher IL6, which was separately associated with AF. Additionally, Large- ASXL1 was associated with increased left ventricle mass index and NT-proBNP (coefficients in Figure 1 ). Conclusion: Large TET2 and ASXL1 but not DNMT3A CHIP is associated with higher IL6 levels, indices of cardiac remodeling, and an increased risk of AF.

Introduction ABBV-383 is a fully human, silenced Fc region IgG4 monoclonal, BCMA × CD3 T-cell-engaging bispecific antibody with 2 BCMA-binding domains and a low-affinity CD3-binding domain. ABBV-383 has shown promising activity following intravenous (IV) infusion every three weeks (Q3W) as monotherapy in patients (pts) with RRMM in an ongoing phase 1 study (D'Souza A, et al. J Clin Oncol. 2022 Nov 1;40(31):3576-3586; Voorhees et al. Blood 2022;140[suppl 1]:4401). Herein, preliminary pharmacokinetic (PK) and immunogenicity results from Q3W regimens of this study and impact of soluble BCMA (sBCMA) on ABBV-383 PK are reported. Methods In the phase 1 study (NCT03933735), adult pts with RRMM, who were previously exposed to ≥3 prior multiple myeloma agents including ≥1 proteasome inhibitor, immunomodulatory drug, and anti-CD38 monoclonal antibody, received Q3W IV infusion at doses 0.025 - 120 mg in escalation (ESC) phase, and at 20 mg, 40 mg, and 60 mg in expansion (EXP) phase. Serial blood samples for PK characterization were collected in cycles 1, 3, and 6 along with sparse/periodic samples throughout the study. Blood samples were also collected throughout the study to detect anti-drug antibodies (ADAs) for immunogenicity characterization and to determine sBCMA concentrations. Appropriate validated assays were used to determine ‘total’ (unbound and bound to sBCMA) and ‘free’ (unbound and partially bound to sBCMA) ABBV-383 serum concentrations, to detect/determine ADAs/titers, and to determine total sBCMA concentrations. ‘Total’ and ‘free’ ABBV-383 serum concentrations from cycles 1, 3, and 6 were used to calculate individual PK parameters (peak concentration [C max], trough concentration [C trough], area under the concentration-time curve [AUC], terminal phase elimination half-life [t 1/2], clearance [CL], and steady state volume of distribution [V ss]) using non-compartmental analysis in Phoenix WinNonlin Version 8.2. Immunogenicity was summarized using ADA/titer data. Impact of baseline/pre-dose sBCMA concentrations on ‘total’ and ‘free’ ABBV-383 PK parameters from cycles 1, 3, and 6 was explored. Results ‘Total’ and ‘free’ ABBV-383 intensive PK data (cycles 1, 3, and 6) were available from 169 and 157 pts, respectively, representing Q3W regimens in ESC and EXP cohorts and ADA data were available from 186 evaluable subjects. Preliminary PK results suggest that ‘total’ and ‘free’ ABBV-383 exhibited approximately dose-proportional PK between 5.4-120 mg Q3W. Estimated t 1/2 for ‘total’ and ‘free’ ABBV-383 ranged between 11-12 days and 8-11 days, respectively, with steady state achieved by cycle 4 in Q3W EXP cohorts. Estimated CL and V ss were consistent with those of a typical IgG4 antibody in humans. Across EXP Q3W regimens, accumulation (cycle 3 vs. cycle 1) was 30%-60% in C max and 60%-177% in C trough for ‘free’ and ‘total’ ABBV-383. ABBV-383 exposures (AUC tau and C trough) of ‘total’ ABBV-383 were 28-74% higher than those representing ‘free’ ABBV-383 among the EXP Q3W regimens with no/minimal observed differences in C max (≤21%). Higher sBCMA levels prior to treatment appeared to reduce ‘free’ ABBV-383 PK exposures with pronounced reduction at lower doses (e.g., 20 mg) and in cycle 1, however the effect diminished in later cycles (cycles 3 and 5). Preliminary assessment of immunogenicity indicated an ADA incidence of 3.8% (7/186) with low titers ranging between 11.7 and 187 titer units and no impact on PK. Conclusions ABBV-383 exhibited a PK profile with relatively long half-life and low incidence of immunogenicity. Overall, sBCMA impacted (reduced) the ‘free’ ABBV-383 (pharmacologically active component) PK exposures only, with pronounced effect at lower doses and the effect is consistent with understanding of treatment dependent sBCMA dynamics over time. Determination of both ‘total’ and ‘free’ ABBV-383 serum concentrations allowed to demonstrate such underlying impact of sBCMA on PK.

Abstract Background Vedolizumab, an anti-α4β7 integrin approved for intravenous (IV) treatment of moderately to severely active ulcerative colitis (UC) and Crohn’s disease (CD), was evaluated as a subcutaneous (SC) formulation in maintenance therapy for UC and CD in phase 3 VISIBLE 1, 2, and open-label extension studies, and recently approved in Europe, Australia, and Canada. Our aim was to evaluate efficacy and safety of IV and SC vedolizumab in clinically relevant UC and CD scenarios. Methods Post hoc data analyses from VISIBLE trials examined: (1) whether baseline characteristics predict clinical response to 2 vs 3 IV vedolizumab induction doses; (2) efficacy and safety of switching during maintenance vedolizumab IV to SC in patients with UC; (3) vedolizumab SC after treatment interruption of 1–46 weeks; (4) increasing dose frequency of vedolizumab SC from every 2 weeks (Q2W) to every week (QW) after disease worsening. Results No baseline characteristics were identified as strong predictors of response to 2 vs 3 vedolizumab infusions. Most patients achieved clinical response after 2 or 3 doses of IV vedolizumab maintained with SC treatment. Clinical remission and response rates were maintained in patients transitioned from maintenance vedolizumab IV to SC treatment. Of patients with UC, ≥75% achieved response following resumption after dose interruption. Escalation to QW dosing resulted in ≥45% of patients regaining response after loss while receiving vedolizumab Q2W. Conclusions Clinical real-world scenarios with vedolizumab SC were reviewed using VISIBLE studies data. Vedolizumab SC provides an additional dosing option for patients with UC and CD.