Abstract Polygenic risk scores (PRSs) may enhance risk stratification for pancreatic ductal adenocarcinoma (PDAC), but existing models vary widely in design, predictive performance, and cross-ancestry transferability. We developed genome-wide PRSs using Bayesian methods (LDpred2 and PRS-CS) and p value thresholding (PRSice-2) and systematically evaluated these alongside 13 published PRSs to identify models with robust predictive performance across ancestries. Using GWAS summary statistics from 7531 cases and 10,631 controls, we derived the PRSs and tested associations in an independent sample of 4508 PDAC cases and 46,189 controls, with adjustment for well-established PDAC risk factors. Among all models, the genome-wide LDpred2-based PRS showed the strongest association with PDAC (OR = 1.57 per standard deviation increase; 95% CI: 1.51–1.62) and significantly improved discrimination beyond established risk factors alone (AUC = 0.74–0.76; p < 0.0001). Importantly, the genome-wide LDpred2 PRS demonstrated consistent associations across African, Admixed American, and European ancestry groups, whereas the best-performing published PRS was associated with PDAC risk only in individuals of European ancestry. These findings support genome-wide PRSs as a promising framework for multi-ancestry risk stratification for PDAC and to inform targeted early detection strategies.

Accurate differentiation of thyroid nodules is crucial for timely diagnosis of thyroid cancer. Most recent studies utilize grayscale ultrasound with deep learning to distinguish benign from malignant thyroid nodules. The goal of this study is to effectively boost the performance of classification of thyroid nodules using multi-modal ultrasound imaging combining B-mode, color Doppler (CD), and shear wave elastography (SWE) within a customized deep learning architecture. This study prospectively included 506 thyroid nodules acquired from 422 subjects. The proposed network integrated a pretrained MobileNetV2 backbone with a shallow head composed of depth-wise separable convolutional layers, attention-based mixed pooling, and a tailored self-attention mechanism—applied for the first time in the context of thyroid nodule classification. To further improve robustness, we introduced a patient-level uncertainty-aware fusion strategy that selectively integrates predictions from each modality based on their validation scheme and achieved a classification accuracy of 0.95, a sensitivity (Sen) of 0.98, an area under the ROC curve (AUC) of 0.97 (95

Long-read nanopore sequencing enables simultaneous detection of germline variation and native DNA base modifications on individual DNA molecules, providing a unique opportunity to investigate allele-specific epigenetic regulation. Here, we performed whole-genome nanopore sequencing on normal and tumor prostate tissues to characterize differential methylation, methylation entropy, and allele-specific methylation (ASM) associated with noncoding genetic variants. Genome-wide analysis identified extensive cancer-associated differentially methylated regions (DMRs), with hypermethylated DMRs significantly enriched near transcription start sites and transcriptional regulatory regions. Integration with transcriptomic datasets revealed strong inverse relationships between promoter methylation and gene expression, while 5-hydroxymethylcytosine (5hmC) levels positively correlated with transcriptional activity across gene bodies. Using fragment-level methylation patterns enabled by long-read sequencing, we further quantified methylation entropy incorporating both 5mCG and 5hmCG states. Cancer-hypermethylated DMRs exhibited markedly reduced entropy, consistent with clonal fixation of methylation states during tumor progression. Entropy profiling across chromatin annotations demonstrated maximal epigenetic heterogeneity at partially modified enhancer-associated regions. To investigate cis-regulatory genetic effects, we developed a simple ASM framework (nanoASM) that can partition sequencing reads by allelic state and identifies allele-specific DMRs directly from long-read data. Compared with conventional population-level mQTL analysis, ASM demonstrated substantially improved statistical efficiency by leveraging within-individual contrasts and reducing sample-level heterogeneity. Although germline single nucleotide polymorphisms (SNPs) were largely shared between normal and tumor tissues, ASM patterns differed substantially, with tumor-associated ASM regions displaying significantly larger genomic span and stronger allelic methylation differences. Comparative analysis with TCGA prostate mQTL and GTEx prostate eQTL datasets demonstrated substantial concordance between ASM directionality and downstream transcriptional effects, particularly for variants located within DMRs and near transcription start sites. At the IRX4 prostate cancer risk locus, ASM identified an androgen-responsive regulatory domain overlapping AR ChIP-seq and H3K27ac peaks, nominating rs6885084 as a candidate functional variant. At the PSCA locus, ASM anchored by rs4736369 was associated with allele-specific methylation, chromatin activation, transcript abundance, and isoform usage. Together, these findings establish nanopore-based ASM analysis as a powerful approach for resolving functional noncoding variants and their regulatory domains they control in prostate cancer.

PURPOSE:To investigate the effectiveness of quantitative biomarkers derived from quantitative high-definition microvasculature imaging (qHDMI) for differentiation of radial scar (RS) and invasive ductal carcinoma (IDC). METHODS:A total of 64 breast lesions from 62 participants were analyzed using breast pathology as the gold standard. Ultrasound data were processed with the qHDMI framework to visualize tumor microvessel networks and extract eight morphological biomarkers. Biomarker distributions were compared between groups using a two-sided Wilcoxon rank-sum test, and ROC AUC values with 95 % confidence intervals (CI) were calculated. P-values were adjusted for multiple testing using the Benjamini-Hochberg FDR method at a 5 % threshold. RESULTS:Seventeen lesions were pathologically confirmed as RS and 47 as IDC. qHDMI biomarkers revealed distinct microvascular differences between the two groups. Overall, IDC lesions showed denser, thicker, and more complex microvessels. Five biomarkers demonstrated statistically significant distribution differences: vessel density (p-value: 0.016, adjusted p-value: 0.0288), number of vessel segments (p-value: 0.0055, adjusted p-value: 0.0288), number of branch points (p-value: 0.0098, adjusted p-value: 0.0288), maximum diameter (p-value: 0.014, adjusted p-value: 0.0288), and microvessel fractal dimension (p-value: 0.018, adjusted p-value: 0.0288). The AUC and CIs for these five biomarkers were 0.70, [0.56, 0.82] for vessel density, 0.73, [0.60, 0.86] for number of vessel segments, 0.71, [0.58, 0.83] for number of branch points, 0.70, [0.57, 0.82] for maximum diameter, and 0.70, [0.54, 0.82] for fractal dimension. CONCLUSION:Distributions of qHDMI-derived biomarkers revealed distinct microvascular structural differences between RS and IDC, suggesting qHDMI may enhance diagnostic accuracy in distinguishing the two.

To address inter-frame motion artifacts in ultrasound quantitative high-definition microvasculature imaging (qHDMI), we introduced a novel deep learning-based motion correction technique. This approach enables the derivation of more accurate quantitative biomarkers from motion-corrected HDMI images, improving the classification of thyroid nodules. Inter-frame motion, often caused by carotid artery pulsation near the thyroid, can degrade image quality and compromise biomarker reliability, potentially leading to misdiagnosis. Our proposed technique compensates for these motion-induced artifacts, preserving the fine vascular structures critical for accurate biomarker extraction. In this study, we utilized the motion-corrected images obtained through this framework to derive the quantitative biomarkers and evaluated their effectiveness in thyroid nodule classification. We segregated the dataset according to the amount of motion into low and high motion containing cases based on the inter-frame correlation values and performed the thyroid nodule classification for the high motion containing cases and the full dataset. A comprehensive analysis of the biomarker distributions obtained after using the corresponding motion-corrected images demonstrates the significant differences between benign and malignant nodule biomarker characteristics compared to the original motion-containing images. Specifically, the bifurcation angle values derived from the quantitative high-definition microvasculature imaging (qHDMI) become more consistent with the usual trend after motion correction. The classification results demonstrated that sensitivity remained unchanged for groups with less motion, while improved by 9.2% for groups with high motion. These findings highlight that motion correction helps in deriving more accurate biomarkers, which improves the overall classification performance.

5-Methylcytosine (5mC) is the most common DNA modification in the human genome. Bisulfite conversion combined with short-read sequencing captures this modification at single-nucleotide resolution but introduces PCR duplication bias and limits co-methylation analysis between distant cytosines. To resolve these limitations, we used nanopore long-read sequencing to profile human methylation and performed long-range co-methylation analysis with native DNA modification information. We analyzed the nanopore demo data in the adaptive sampling sequencing targeting the CpG islands and applied the linkage disequilibrium (LD) R2 to identified methylation haplotype blocks (MHBs). We found that the cancer genome exhibited significantly smaller MHBs, higher CpG density, and a lower methylation LD R2 value compared to normal cells. Additionally, we demonstrated the superiority of long-read sequencing in capturing large MHBs compared with short-read sequencing. By profiling the methylation changes near the JASPAR motif and actual chromatin immunoprecipitation sequencing (ChIP-seq) peaks, we also studied the epigenetic changes related to protein binding. Based on adaptive sampling technology, we conducted nanopore sequencing targeting regions with methylation quantitative trait loci (mQTLs) and genome-wide association study (GWAS) risk variants in the 22Rv1 cell line. After analyses, we inspected the closest haplotype-specific methylated region near the variant and identified allele-specific methylated regions with allele-specific accessibility signals in the ATAC-seq data. This study demonstrates the feasibility of nanopore sequencing for methylome profiling while preserving haplotype information, offering an innovative approach to elucidate the epigenetic changes driven by noncoding variants in the human genome.

BACKGROUND:This research aims to evaluate wave speed and viscoelasticity of ocular tissues including the optic nerve and optic nerve head of human eyes between normal tension glaucoma patients and healthy controls by using vibro-elastography techniques. METHODS:Participants included 12 patients and 12 controls. Wave speed was measured at the optic nerve and optic nerve head in each subject and viscoelasticity was estimated by using Voigt model. Wave speed and viscoelasticity of the optic nerve and optic nerve head were compared between patients and controls by linear mixed models via a restricted maximum likelihood method. The correlation between intraocular pressure and wave speed, elasticity, and viscosity of patients was performed using the Pearson correlation coefficient. FINDINGS:Significant differences in wave speed (p < 0.0005), elasticity (p = 0.0001) and viscosity p < 0.0001) between patients and controls at the optic nerve head. There was a moderate negative correlation (r = -0.50, p < 0.05) between wave speed and elasticity and intraocular pressure at the optic nerve of patients but no correlation (p > 0.05) between wave speed, elasticity, and viscosity and intraocular pressure at the optic nerve head of patients. No significant difference and correlation among wave speed, elasticity, and viscosity vs. intraocular pressure of the control group at the optic nerve and optic nerve head. INTERPRETATION:Ultrasound vibro-elastography is useful for noninvasive measurement of viscoelasticity of ocular structures. The glaucoma patient is associated with biomechanical property changes in the optic nerve and optic nerve head, suggesting another way to assess glaucoma focusing on the optic nerve and optic nerve head.

PURPOSE:To propose a multi-parametric ultrasound imaging-based deep learning method for accurately classifying metastatic and non-metastatic axillary lymph nodes in breast cancer patients. METHODS:The proposed method integrates the conventional ultrasound B-mode imaging with shear wave elastography and color Doppler images of 174 patients to train a transfer learning-based network comprising pretrained MobileNetv2 with a custom shallow head consisting of a convolutional neural network with mixed pooling, weighted sum mixed pooling and squeeze-and-excite attention mechanisms for the first time in the context of ALN classification. RESULTS:The proposed method was evaluated using five-fold cross-validation, achieving a mean classification accuracy of 0.91, specificity of 0.91, sensitivity of 0.93, F1 score of 0.93, area under the precision-recall curve of 0.94, and a cross-validated AUC (cvAUC) of 0.92. A network ablation study confirmed the robustness of the model, with relatively narrow 95% confidence intervals (CIs) for cvAUC. Comparative analysis showed that the proposed network (Acc: 0.91) outperformed state-of-the-art deep learning models (Acc: 0.67-0.88) for ALN classification and exhibited narrower CIs, highlighting its relative stability. Additionally, results demonstrated that multi-parametric imaging significantly enhanced classification performance, reducing the 95% CI width by nearly half compared to uni-parametric data, further supporting the method's robustness and reliability. CONCLUSION:The integration of multi-parametric ultrasound imaging with deep learning network can remarkably improve the classification of metastatic and non-metastatic ALNs in breast cancer patients.

Key PointsCirculating proteins independently associated with kidney stone diagnosis are related to kidney stone matrix.Lower circulating uromodulin and higher scavenger receptor cysteine-rich domain-containing group B protein levels were associated with higher kidney stone risk, whereas kidney stone presence elevates plasma matrix metalloproteinase 7 level.The protective effect of uromodulin against kidney stones was independent of kidney function.BackgroundKidney stones are increasingly recognized as a systemic disorder with a high global prevalence. However, large proteomics studies are lacking.MethodsAn individual-level proteomics study with rigorous adjustments was performed on 35,331 UK Biobank participants to uncover the independent associations between 2922 circulating proteins and prevalent kidney stone disease. Mendelian randomization analysis was used to assess causal relationships. Findings were validated using genomic data from the Mayo Clinic Biobank (N=43,744), concentration-response analysis, transcriptomics analyses, and additional genome-wide association studies.ResultsNine plasma proteins were independently associated with a kidney stone diagnosis, including reduced uromodulin (UMOD; beta, -0.10; 95% confidence interval [CI], -0.15 to -0.05) and elevated scavenger receptor cysteine-rich domain-containing group B protein (SSC4D; beta, 0.28; 95% CI, 0.17 to 0.38) and were enriched in extracellular matrix pathways. Mendelian randomization analysis revealed that the presence of kidney stone contributed to elevated levels of matrix metalloproteinase 7. Conversely, lower plasma UMOD (odds ratio [OR], 0.93; 95% CI, 0.90 to 0.97) and higher plasma SSC4D (OR, 1.10; 95% CI, 1.02 to 1.18) were associated with kidney stone risk. These associations were consistently replicated in the Mayo Clinic Biobank dataset (UMOD; OR, 0.92; 95% CI, 0.86 to 0.98; SSC4D; OR, 1.13; 95% CI, 1.01 to 1.27) and further validated by a concentration-response analysis. Single-nucleus RNA sequencing and quantitative trait loci analyses confirmed consistent associations between thick ascending limb UMOD expression and stone former status and with blood and urine UMOD concentrations. Genome-wide association study analysis, adjusted for eGFR, suggested that the protective role of UMOD against kidney stones was independent of kidney function.ConclusionsThis study highlights significant associations between concentrations of specific blood proteins and a history of kidney stones. Several implicated proteins are related to kidney stone matrix, with UMOD independently associated with lower risk and SSC4D with higher risk of kidney stones.

OBJECTIVE:Investigate shared genetic changes that contribute to the association of kidney stones with cardiometabolic comorbidities. PATIENTS AND METHODS:Genome-wide association studies (GWAS) and meta-analyses were performed using the UK Biobank, Mayo Clinic Biobank, and FinnGen, complemented by secondary genetic association analyses. RESULTS:Genetic correlation analyses confirmed modest but significant positive associations between kidney stones and hypertension, obesity, and type 2 diabetes (rg=0.14 ∼ 0.19, all P<.001). Single nucleotide variants near FTO (for expansion of gene symbols, use search tool at www.genenames.org) were identified as shared risk loci across all conditions. Mendelian randomization analyses demonstrated that higher body mass index (BMI) and waist-hip ratio (WHR) significantly increased the risk of kidney stones, type 2 diabetes, and hypertension. Subsequent GWAS analyses adjusting for BMI or WHR revealed a slightly reduced yet significant genetic correlation between kidney stones and type 2 diabetes (BMI-adjusted: rg=0.18; 95% CI, 0.07 to 0.28; P=.001; WHR-adjusted: rg=0.16; 95% CI, 0.06 to 0.25; P<.001), whereas the association of kidney stones with hypertension was nearly eliminated. A multi-trait analysis of GWAS identified one potential shared risk locus between BMI-adjusted kidney stones and type 2 diabetes on chromosome 6 near LINC02537. Further analyses indicated this locus is associated with thyroid-stimulating hormone and free thyroxine levels. CONCLUSION:Our study highlights a critical role for shared obesity-related genetic changes in the comorbid relationship between kidney stones and cardiometabolic conditions. There are additional genetic features related to thyroid function that are shared between kidney stones and type 2 diabetes, which appear to be independent of obesity.

Microvessel orientations are known to differ between benign and malignant tumors. This article introduces novel orientation-based quantitative biomarkers for contrast-free ultrasound microvasculature imaging to distinguish malignant from benign breast lesions. The proposed biomarkers were computed in both polar and Cartesian coordinates using images acquired by a high-definition microvasculature imaging technique. Seven biomarkers were derived based on the histogram, gradient angle, angle of penetration, and penetration factor of the microvessel. These biomarkers are first evaluated using simulated microvessel images and subsequently validated in in vivo studies on human breast masses. The new biomarkers demonstrated statistical significance in differentiating benign from malignant breast masses. In the in vivo study, the area under the receiver operating characteristic curve (AUC) for the proposed biomarkers was 0.91 (95% confidence interval [CI]: 0.86, 0.97). When the Breast Imaging Reporting and Data System (BI-RADS) score was included in the classification model, the AUC improved to 0.97 (95% CI: 0.91,1.00). These orientation-based biomarkers show promise in enhancing the diagnostic performance of ultrasound for classification of breast masses.

ABSTRACT Enzalutamide and abiraterone are hormonal treatments that improve survival in metastatic castration‐resistant prostate cancer. Identifying genetic variants associated with the clearance of these drugs may aid in improved dosing and outcomes. We performed genetic association studies of enzalutamide and abiraterone oral clearance in the Alliance A031201 clinical trial. Genome‐wide genotyping was performed with the primary analysis limited to European‐descent participants. Pharmacogene metabolic phenotypes were estimated using PyPGx and Stargazer. Associations of metabolic activity groups for CYP3A4, CYP3A5, CYP2C19 and SLCO1B1 with enzalutamide clearance (N = 706) and CYP3A4, SLCO2B1 and UGT1A4 with abiraterone clearance (N = 323) were tested by linear regression. Targeted SNP associations were assessed for abiraterone clearance at loci proximal to major metabolizing genes. Full genome‐wide association studies were performed for both sets of clearance values. No significant associations were identified between metabolic phenotypes and enzalutamide or abiraterone oral clearance SNPs in the SULT2A1 5′ flanking region were significantly associated with lower abiraterone clearance, (rs296373, minor allele frequency = 0.15, β = −0.457, p = 3.2E‐06). Liver protein and liver and adrenal gland gene expression QTL databases indicated significantly lower SULT2A1 expression patterns for individuals carrying associated alleles, likely explaining the lower abiraterone oral clearance. CYP2C8*3 was associated with higher enzalutamide clearance (p = 0.012), but this was not significant after correction for multiple testing. This study is the first to identify the genetic association of SULT2A1, known to be involved in the metabolism of steroids in the liver and adrenal glands, with abiraterone clearance. Genetic variation in SULT2A1 may be useful to inform personalized dosing of abiraterone. ClinicalTrials.gov Identifier: NCT01949337

Numerous deep learning (DL) and foundational models (FMs) designed for digital pathology analysis employ advanced feature extraction techniques from digitized hematoxylin and eosin (H E) slides obtained from tissue sections. These extracted features may serve as the input for complex models enabling automated classification tasks, slide searching, and other various medical inferences. Subtle variations in morphology observed in distinct cross sections of the same tissue sample may introduce variability that may be of little to no clinical relevance. The reproducibility of features derived from DL models and FMs on ostensibly identical tissue cross sections remains largely unexplored. In this study, we aimed to characterize the reproducibility of features extracted from three sequential cross sections of 50 independent normal prostate samples. Slides were sectioned at approximately 50 μm intervals and we extracted cell-type specific counts, proportions, and spatial features per slide using DL models, along with feature embeddings from popular digital pathology FMs. Reproducibility of the features across sequential cross sections of tissues was assessed using intra-class correlation coefficient (ICC) for DL model features, while extracted FM embedding distributions were evaluated via maximum mean discrepancy (MMD), and Wasserstein’s distance. The median [IQR] ICC estimates for cell-type specific proportion and spatial features from the cross sections was 0.838 [0.764, 0.897], and 0.873 [0.809, 0.923] respectively. The bootstrapped mean differences of ICC estimates for the features were significantly lower (p-value < 0.05) for distant cross sections compared to the two neighboring cross sections. The feature embeddings derived from state-of-the-art digital pathology FMs showed an overall high agreement with the median [IQR] MMD of 0.06 [0.016, 0.169], and median [IQR] Wasserstein’s distance of 0.021 [0.011, 0.038]. These findings demonstrate that the features derived from multiple cross sections of the same tissue exhibit overall high reproducibility, while also illustrating the expected increase in intra-sample variability as a function of physical distance between sections.

BACKGROUND:Kidney aging is characterized by a loss of glomeruli, predominately in the superficial cortex, with a resultant decline in glomerular filtration rate and an increased risk of various kidney-related diseases. The early molecular alterations in glomeruli associated with the aging process are not well studied. METHODS:We combined laser capture microdissection and mass spectrometry-based unbiased proteomic analysis of non-sclerosed, non-ischemic glomeruli in the superficial cortex from young and old adults who underwent a radical nephrectomy for a tumor to understand the age-related molecular changes in glomeruli. 24 young and 30 old adults were used for the discovery dataset and the significant differentially expressed proteins were further validated using an independent set comprising 6 young and 8 old adults. RESULTS:Kidneys from older adults had lower eGFR, less kidney parenchyma on CT imaging, and more glomerulosclerosis and arteriosclerosis on histology. Quantitative proteomic analysis of non-sclerosed, non-ischemic glomeruli identified increased expression of TIMP3, GPC6, SNCG, APOA4 and NT5E in old adults that were further validated in an independent set. Pathway analysis indicated that proteins with increased expression in old adults were enriched in mitochondrial translational processes, aerobic respiration, and TCA cycle, whereas proteins with decreased expression in old adults were enriched in mRNA splicing, mRNA processing, and nonsense mediated decay. Further, Spearman correlation of validated differentially expressed proteins did not show any significant correlation with the kidney pathology independent of age group. CONCLUSIONS:Overall, this study identified proteins that are specifically associated with the aging process in otherwise normal-appearing glomeruli.

BACKGROUND:Familial hypercholesterolemia (FH) is an autosomal dominant genetic disorder that increases risk for premature coronary artery disease and has accessible and effective interventions. The Dutch lipid clinic network is currently the most used diagnostic criterion; however, genetic sequencing provides a definitive diagnosis of FH. The goals of this study were to determine whether germline genetic screening using exome sequencing could be used to efficiently identify individuals who were genotype positive for FH. METHODS:Participants were recruited from 3 geographically and racially diverse sites in the United States (Rochester, MN; Phoenix, AZ; and Jacksonville, FL). Participants underwent Exome+ sequencing (dba Helix, San Mateo, CA) and return of results for specific genetic findings in APOB, LDLR, or PCSK9. A chart review was performed to collect demographics, personal, and family cardiovascular history. RESULTS:At the time of the study, 84 413 participants were enrolled in the Tapestry study. Annotation and interpretation of all variants in genes for FH resulted in the identification of 419 likely pathogenic and pathogenic variants (prevalence, 0.50%), which included 116 APOB, 298 LDLR, and 5 PCSK9. Sixty-six percent were female, the mean body mass index was 27.3, with 12.3% reporting a history of diabetes. Hypertriglyceridemia (≥150 mg/dL) was present in 39.5% and reduced HDL (<50 mg/dL) was present in 56.7% of patients. 27.5% of patients were not on cholesterol-lowering medications, and only 10% of FH carriers were at goal low-density lipoprotein cholesterol levels. A history of coronary artery disease was reported in 22.4% of the cohort. Nearly 90% of these participants were newly diagnosed carriers of FH. Only 30.8% of confirmed genetic diagnoses of FH satisfied clinical (Dutch lipid clinic network) criteria for a diagnosis. CONCLUSIONS:Our results emphasize the need for wider utilization of germline genetic sequencing for enhanced screening and detection of individuals who have familial hypercholesterolemia. REGISTRATION:URL: https://www.clinicaltrials.gov; Unique identifier: NCT05212428.

The preservation of tissue architecture and morphology in formalin-fixed paraffin-embedded (FFPE) tissues enables spatial resolution at the cellular and sub-cellular levels. Laser capture microdissection (LCM) combined with liquid chromatography tandem mass spectrometry analysis permits collection of tissue areas with spatial context for proteome profiling from FFPE slides. In this study, we performed proteome profiling of non-diseased renal tubulointerstitial tissue in a cohort of young (< 40 years) and old (> 70 years) individuals with the goal of spatially correlating the histomorphology to the proteomic profile. To perform in-depth characterization of renal tubulointerstitium and to identify renal aging-associated proteins, a multiplexing strategy using tandem mass tags (TMT) was employed, resulting in the quantitation of 7,355 proteins. Our approach allowed for identification of proteins with low abundance such as fibrocystin and ninein-like protein. Notably, 162 solute carrier proteins from 47 solute carrier families were identified, which were enriched in proximal and distal tubule cells. Finally, we discovered a proteomic signature associated with renal aging, which includes metalloproteinase inhibitor 3, nicotinamide N-methyltransferase, matrix metallopeptidase 7, phenazine biosynthesis-like domain-containing protein and solute carrier family 23 member 1. Overall, our study demonstrates the power of LCM combined with proteomics to leverage archived FFPE tissue samples for investigating proteomic alterations in the renal tubulointerstitium with age at a high depth of proteome coverage.

Background and purpose: Even though glioblastoma (GB) and brain metastases (BM) can be differentiated using radiomics, it remains unclear if the model performance may vary based on the contrast-enhanced sequence used. Our aim was to evaluate the radiomics-based model performance for differentiation between GB and brain metastases BM using MPRAGE and volumetric interpolated breath-hold examination (VIBE) T1-contrast-enhanced sequence. Materials and methods: T1 contrast-enhanced (T1-CE) MPRAGE and VIBE sequences acquired in 108 patients (31 GBs and 77 BM) during the same MRI session were retrospectively evaluated. After standardized image preprocessing and segmentation, radiomics features were extracted from necrotic and enhancing tumor components. Pearson correlation analysis of radiomics features from tumor subcomponents was also performed. A total of 90 machine learning pipelines were evaluated using a 5-fold cross-validation. Performance was measured by mean area under the curve (AUC)-receiver operating characteristic (ROC), log loss, and Brier scores. Results: A feature-wise comparison showed that the radiomics features between sequences were strongly correlated, with the highest correlation for shape-based features. The mean AUC across the top 10 pipelines ranged between 0.851 and 0.890 with T1-CE MPRAGE and between 0.869 and 0.907 with the T1-CE VIBE sequence. The top-performing models for the MPRAGE sequence commonly used support vector machines, while those for the VIBE sequence used either support vector machines or random forest. Common feature-reduction methods for top-performing models included linear combination filter and least absolute shrinkage and selection operator for both sequences. For the same machine learning feature-reduction pipeline, model performances were comparable (AUC-ROC difference range, -0.078-0.046). Conclusions: Radiomics features derived from T1-CE MPRAGE and VIBE sequences are strongly correlated and may have similar overall classification performance for differentiating GB from BM.