eBioMedicine

BackgroundSerum mitochondrial DNA (mtDNA) fragments act as proinflammatory damage-associated molecular patterns (DAMPs), and have been linked to outcomes in critical illness. However, their prognostic value remains uncertain, possibly due to confounding nuclear mitochondrial insertions (NUMTs) which obscure both quantitation and variant detection. MethodsUsing a targeted deep sequencing and bioinformatics workflow, we created filtering strategies to minimize NUMT-related artifacts. To evaluate the method, we performed a secondary analysis of serum samples collected from NCT00976833, a study of acute respiratory failure patients. By modeling DNA insert size distributions, we excluded likely NUMT-derived DNA fragments based on their size, improving the accuracy of mtDNA DAMP fragmentomic analysis. To improve variant detection, we introduced a novel "read mismatch percentage" metric to identify NUMT-induced chimeric read pairs, enabling identification of mtDNA variants. ResultsMean NUMT-depleted, but not raw, mtDNA insert size was lower in non-survivors. Short DNA inserts (<150 bp) displayed little NUMT contamination, and their abundance and size correlated with mortality more strongly than total mtDNA abundance. Sequence variants were called and some associated with survival and post-acute quality of life. Variant m.1,719G>A, found in small humanin-like 3 (MT-SHLP3), associated with survival. Other variants associated with overall poor outcome (non-survival or poor QoL). Two noncoding variants previously associated with low VO2 max and coronary artery disease (m.295C>T and m.462C>T) also associated with poor outcome in the present study. Two MT-ND5 variants m.13,708G>A (a missense variant previously implicated in kidney dysfunction) and m.12,612A>G (a synonymous variant previously associated with coronary artery disease) also associated with poor overall outcome. ConclusionsOur results addressed limitations of standard qPCR-based methods for the study of mtDNA DAMPs. Beyond addressing confounding NUMT, the method identified fragmentomic and variant associations overlooked by qPCR. Cell-free DNA fragmentomic and variant information are well-established biomarkers for cancer, and this method could facilitate similar patient-specific biomarkers in the context of critical illness. The method is composed of commercially available reagents and open source software, which could additionally promote adoption and reproducibility.

BackgroundThe retina is considered a unique window to systemic health, but their biological link remains unknown. MethodsA total of 93,838 UK Biobank participants with metabolomics data were included in the study. Plasma metabolites associated with GCIPLT were identified in 7,824 participants who also underwent retinal optical coherence tomography; prospective associations of GCIPLT-associated metabolites with 12-year risk of mortality and major age-related diseases were assessed in 86,014 participants. The primary outcomes included all- and specific-cause mortality. The secondary outcomes included incident type 2 diabetes mellitus (T2DM), obstructive sleep apnea/hypopnea syndrome (OSAHS), myocardial infarction (MI), heart failure, ischemic stroke, and dementia. C-statistics and net reclassification indexes (NRIs) were calculated to evaluate the added predictive value of GCIPLT metabolites. Calibration was assessed using calibration plots. FindingsSixteen metabolomic signatures were associated with GCIPLT (P< 0.009 [Bonferroni-corrected threshold]), and most were associated with the future risk of mortality and age-related diseases. The constructed meta-GCIPLT scores distinguished well between patients with high and low risks of mortality and morbidity, showing predictive values higher than or comparable to those of traditional risk factors (C-statistics: 0.780[0.771-0.788], T2DM; 0.725[0.707-0.743], OSAHS; 0.711[0.695-0.726], MI; 0.685[0.662-0.707], cardiovascular mortality; 0.657[0.640-0.674], heart failure; 0.638[0.636-0.660], other mortality; 0.630[0.618-0.642], all-cause mortality; 0.620[0.598-0.643], dementia; 0.614[0.593-0.634], stroke; and 0.601[0.585-0.617], cancer mortality). The NRIs confirmed the inclusion of GCIPLT metabolomic signatures to the models based on traditional risk factors resulted in significant improvements in model performance (5.18%, T2DM [P=3.86E-11]; 4.43%, dementia [P=0.003]; 4.20%, cardiovascular mortality [P=6.04E-04]; 3.73%, MI [P=1.72E-07]; 2.93%, OSAHS [P=3.13E-05]; 2.39%, all-cause mortality [P=3.89E-05]; 2.33%, stroke [P=0.049]; 2.09%, cancer mortality [P=0.039]; and 1.59%, heart failure [P=2.72E-083.07E-04]). Calibration plots showed excellent calibration between predicted risk and actual incidence in the new models. InterpretationGCIPLT-associated plasma metabolites captured the residual risk for mortality and major systemic diseases not quantified by traditional risk factors in the general population. Incorporating GCIPLT metabolomic signatures into prediction models may assist in screening for future risks of these health outcomes. FundingNational Natural Science Foundation (China). Research in contextO_ST_ABSEvidence before this studyC_ST_ABSRecent studies have recognized that retinal measurements can indicate an accelerated risk of aging and multiple systemic diseases preceding clinical symptoms and signs. Despite these insights, it remains unknown how retinal alterations are biologically linked to systemic health. Added value of this studyUsing the UK Biobank, we identified ganglion cell-inner plexiform layer thickness (GCIPLT) metabolomic signatures, and revealed their association with the risk of all- and specific-cause mortality and six age related diseases: type 2 diabetes, dementia, stroke, myocardial infarction, heart failure, and obstructive sleep apnea/hypopnea syndrome. The meta-GCIPLT score significantly improved the discriminative power of the predictive models for theses health outcomes based on conventional risk factors. Implications of all the available evidenceGCIPLT-associated plasma metabolites have the potential to capture the residual risk of systemic diseases and mortality not quantified by traditional risk factors. Incorporating GCIPLT metabolomic signatures into prediction models may assist in screening for future risks of these health outcomes. Since metabolism is a modifiable risk factor that can be treated medically, the future holds promise for the development of new strategies that reverse or interrupt the onset of these diseases by modifying metabolic factors.

PurposeTo test whether internal memory states from a medical founda-tional segmentation model can serve as compact, mask-aware embeddings for predicting progression-free survival (PFS) in multiple myeloma (MM) from whole-body [18F]FDG PET/CT, and how late fusion of PET, CT, and clinical data enhances prognostic performance. MethodsWe analyzed 227 newly diagnosed MM patients with PET/CT and clinical data. For two regions of interest (spine-dilated and full skeleton), we prompted MedSAM2 slice-wise using mask-derived bounding boxes and cached the final spatio-temporal memory tensor per modality. We compared two downsampling strategy to obtain per-study embeddings: channelxmemory averaging with a small CNN head, and depth-attention pooling. PET and CT embeddings were combined by late fusion and passed to a DeepSurv head. We evaluated image-only and multimodal (image+clinical) models with stratified 5-fold cross-validation. The primary endpoint was Harrells c-index (mean {+/-} SE across folds). ResultsImage-only models using the averaging downsampler achieved up to 0.659 {+/-} 0.015 c-index (PET, spine-dilated), comparable to baseline ra-diomics results. Multimodal models improved discrimination to 0.710{+/-}0.032 (CT, spine-dilated), with similar performance for other PET/CT+clinical variants (0.703-0.710), improving clinical-only baselines [~] 6.5%. Averaging consistently outperformed depth-attention; concatenation and gated fusion performed comparably. PET outperformed CT within the same mask in image-only settings. ConclusionMask-aware memory embeddings extracted from a founda-tional segmentation model provide effective, data-efficient imaging biomark-ers for MM PFS and, when fused with routine clinical covariates, significantly improve risk stratification over clinical-only or radiomics baselines. This of-fers a practical path to prognostic modeling on small medical cohorts without feature design.

A notable increase in severe cases of COVID-19, with significant hospitalizations due to the emergence and spread of JN.1 was observed worldwide in late 2023 and early 2024. During the study period (November 2022-January 2024), 56 JN.1- and 126 XBB-infected patients were prospectively enrolled in 40 French intensive care units. JN.1-infected patients were more likely to be obese (35.7% vs 20.8%; p=0.033) and less frequently immunosuppressed than others (20.4% vs 41.4%; p=0.010). JN.1-infected patients required invasive mechanical ventilation support in 29.1%, 87.5% of them received dexamethasone, 14.5% tocilizumab and none received monoclonal antibodies. Day-28 mortality of JN.1-infected patients was 14.6%.

Withdrawal statementThe authors have withdrawn their manuscript owing to altered the findings and conclusions related to complication data. The changes in results were due to further scrutiny of the datasets, and led to the removal of 3 patients due to incomplete data. This refinement led to updated results which changed the conclusion of the complication data. Therefore, the authors do not wish this work to be cited as reference for the project. If you have any questions, please contact the corresponding author.

DisclaimerThis manuscript has been withdrawn by the corresponding author, as it was submitted and made public in bioRxiv without knowledge and the full consent of all the authors listed. The co-authors are therefore not responsible for the contents of this manuscript. For this reason, the submitting author has chosen to withdraw this preprint as a precautionary measure and would like to state that this work should not be cited as a reference. For any related questions that may arise, please contact the corresponding author.Competing Interest StatementThe authors have declared no competing interest.

BackgroundAdverse gastric symptoms persist in up to 20% of fundoplication surgeries completed for gastroesophageal reflux disease, causing significant morbidity, and driving the need for revisional procedures. Non-invasive techniques to assess the mechanisms of persistent postoperative symptoms are lacking. We aimed to investigate gastric myoelectrical abnormalities and symptoms in patients after fundoplication using a novel non-invasive body surface gastric mapping (BSGM) device. MethodsPatients with previous fundoplication surgery and ongoing significant gastroduodenal symptoms, and matched controls were included. BSGM using Gastric Alimetry (Alimetry, New Zealand) was employed, consisting of a high resolution 64-channel array, validated symptom-logging App, and wearable reader. Results16 patients with significant chronic symptoms post-fundoplication were recruited, with 16 matched controls. Overall, 6/16 (37.5%) patients showed significant spectral abnormalities defined by unstable gastric myoelectrical activity (n = 2), abnormally high gastric frequencies (n = 3) or high gastric amplitudes (n = 1). Those with spectral abnormalities had higher Patient Assessment of Upper Gastrointestinal Disorders-Symptom Severity Index scores (3.2 [2.8 to 3.6] vs 2.3 [2.2 to 2.8]; p =0.024). 7/16 patients (43.8%) had Gastric Alimetry tests suggestive of gut-brain axis contributions, and without myoelectrical dysfunction. Increasing Principal Gastric Frequency deviation, and decreasing Rhythm Index were associated with symptom severity (r>0.40, p<0.05). ConclusionA significant number of patients with persistent post-fundoplication symptoms display abnormal gastric function on Gastric Alimetry testing, which correlates with symptom severity. These findings advance the pathophysiological understanding of post-fundoplication disorders which may inform diagnosis and patient selection for medical therapy and revisional surgery.

BackgroundPost-COVID-19 condition (PCC) affects millions globally, presenting as persistent multisystem symptoms. Despite various hypotheses, the biological mechanisms underlying PCC remain unclear. Previous studies have linked PCC to alterations in the DNA methylome of blood immune cells, but the effects on lung cells over time remain unknown. MethodsPatients (n=13) with persistent symptoms following COVID-19 in 2020-2021 donated blood and sputum samples at inclusion and after one year. Symptom and physiological testing data were collected concurrently. DNA methylation (DNAm) profiles were analysed longitudinally and compared to pre-pandemic DNAm data from healthy controls. ResultsWhile peripheral blood mononuclear cells (PBMCs) showed no significant changes, longitudinal DNAm changes were observed in neutrophil- and macrophage-enriched fractions. The changes were significantly associated with symptoms and physiological measures. Pathway analysis showed enrichment for cellular processes involved in cardiac function. ConclusionsWe identified longitudinal DNAm changes in lung immune cells associated with pathways linked to cardiac function. These changes correlate with symptom burden and heart and lung metrics. The results suggest potential disease mechanisms and aid the development of diagnostic tools.

BackgroundOrgan fibrosis caused by the presence of excessive extracellular matrix (ECM) is strongly related to mortality. Urinary peptide signatures were reported predictive of death in SARS-CoV-2 and chronic kidney disease. Such signatures were composed for 68% of collagen fragments. In this study, we examined whether an exclusively collagen-based urinary peptide model, potentially representing organ fibrosis, could predict mortality in patients with critical and non-critical conditions. MethodsUrinary proteomic data from 1,012 patients infected with SARS-CoV-2 were considered to evaluate the association of collagen peptide levels with short term mortality. Additional datasets from 9,193 patients were used for validation, including 1,719 patients sampled at intensive care unit (ICU) admission and 7,474 patients with other diseases (outside the ICU) were extracted from the Human Urinary Proteome Database. FindingsA collagen peptide-based model based on 210 peptides (COL210) predicting mortality was developed for patients after SARS-CoV-2 infection. This model was validated in patients in (ICUHR: 2{middle dot}64; 1{middle dot}71-4{middle dot}10; p<0{middle dot}001) and outside the ICU (Non-ICUHR: 2{middle dot}16 95% CI: 1{middle dot}47-3{middle dot}17; p<0{middle dot}001), showing strong associations to mortality regardless underlying conditions. InterpretationThis study demonstrates a link between the presence of ECM fragments in urine, specifically collagens, and increased mortality risk. The availability of such a non-invasive collagen-based predictor of mortality may serve as basis for proteomics guided targeted intervention. FundingThis study was funded by "DisCo-I" (HORIZON-MSCA; 101072828), "SIGNAL" (BMBF; 01KU2307, FWF; project number I 6471 and Grant-DOI 10.55776/I6471 and ANR-22-PERM-0002-06), "UriCov" (BMG,2523FSB114), by "Accurate-CVD" (BMWK), by UPTAKE (BMBF; 01EK2105A-C) and MULTIR (101136926) Research in ContextO_ST_ABSEvidence before this studyC_ST_ABSFibrosis, marked by excessive extracellular matrix (ECM) deposition, is a key factor in chronic diseases and organ failure and is correlated with increased mortality in conditions such as idiopathic pulmonary fibrosis (IPF), chronic kidney disease (CKD), liver disease, cardiovascular disease (CVD), and cancer. A literature search conducted on 10/12/2024 using the MeSH terms "collagen" AND "fibrosis" AND "mortality" OR "death" OR "failure", revealed a shift from static histological assessments of ECM deposition to a more dynamic approach based on the assessment of biomarkers of collagen turnover within specific fibrotic conditions, which offer real-time and potentially actionable insights into disease progression and predict adverse outcomes, including mortality. Key biomarkers include products of collagen synthesis (e.g., pro-peptides of type III, V, and VI collagen), but also of collagen degradation. These biomarkers have individually shown correlations with fibrosis severity and mortality, with links to disease progression and survival in IPF, CKD, CVD, and cancer. In heart failure, urinary peptides from collagen -1 (I) chain predict adverse events and mortality. Collagen -1 (XXIV) chain was also among the circulating plasma proteins that were related with transplant-free survival of patients with IPF. Added value of this studyThis work further adds to the growing relevance of collagen turnover in the context of mortality across fibrotic diseases, in an effort to generalize evidence for collagen degradation markers, independent from the underlying pathological condition. It additionally explores integration of collagen degradation fragments into risk models, potentially improving predictive accuracy and advancing precision medicine. Implications of all the available evidenceOur study demonstrates that urinary collagen degradation markers integrated into a machine learning model (COL210) can identify "vulnerable" individuals from different fibrotic backgrounds, offering potential for improved prognostication and targeted interventions.

Purpose50-60% of neuromuscular-disease patients remain undiagnosed even after extensive genetic testing that hinders precision-medicine/clinical-trial-enrollment. Importantly, those with DNA-based molecular diagnosis often remain without known molecular mechanism driving different degrees of disease severity that hinders patient stratification and trial-readiness. These are due to: a) clinical-genetic-heterogeneity (eg: limb-girdle-muscular-dystrophies(LGMDs)>30-subtypes); b) high-prevalence of variants-of-uncertain-significance (VUSs); (c) unresolved genotype-phenotype-correlations for patient stratification, and (d) lack of minimally-invasive biomarker-driven-assays. We therefore implemented a combinatorial phenotype-driven blood-biomarker functional-genomics approach to enhance diagnostics and trial-readiness by elucidating disease mechanisms of a neuromuscular-disease patient-cohort clinically-suspected of Dysferlinopathy/related-LGMD, the second-most-prevalent LGMD in the US. MethodsWe used CD14+monocyte protein-expression-assay on 364 Dysferlinopathy/related-LGMD-suspected patient-cohort without complete molecular-diagnosis or genotype-phenotype correlation; and then combined with blood-based targeted-transcriptome-sequencing (RNA-Seq) with tiered-analytical-algorithm correlating with clinical-measurements for a subset of patients. ResultsOur combinatorial-approach significantly increased the diagnostic-yield from 25% (N=326; 18%-27%; 95%CI) to 82% (N=38; 69.08% to 84.92%; 95% CI) by combining monocyte-assay with enhanced-RNA-Seq-analysis and clinical-correlation, following ACMG-AMP-guidelines. The tiered-analytical-approach detected aberrant-splicing, allele-expression-imbalance, nonsense-mediated-decay, and compound-heterozygosity without parental/offspring-DNA-testing, leading to VUS-reclassifications, identification of variant-pathomechanisms, and enhanced genotype-phenotype resolution including those with carrier-range Dysferlin-protein-expression and milder-symptoms, allowing patient-stratification for better trial-readiness. We identified uniform-distribution of pathogenic-variants across DYSF-gene-domains without any hotspot suggesting the relevance of upcoming gene-(full-DYSF-cDNA)-therapy trials. ConclusionOur results show the relevance of using a clinically-driven multi-tiered-approach utilizing a minimally-invasive biomarker-functional-genomic platform for precision-medicine-diagnostics, trial-recruitment/monitoring, elucidating pathogenic-mechanisms for patient stratification to enhance better trial outcomes, which in turn, will guide more rational use of current-therapeutics and development of novel-interventions for neuromuscular-disorders, and applicable to other genetic-disorders.

BackgroundThe systemic immune-inflammation index (SII) reflects the relationship between tumor-promoting inflammation and anti-tumor immunity in various solid malignancies, but the role of SII in Gallbladder cancer (GBC) has yet to be established. The aim of the systematic review and meta-analysis was to clarify the prognostic value of preoperative SII in GBC patients undergoing resection. MethodsThis systematic review and meta-analysis was performed following PRISMA 2020 checklist and we searched PubMed, Embase, Web of Science and Scopus from inception to November 1, 2025. Original studies with English language enrolled adults with resectable GBC undergoing surgical resection and reported preoperative SII with overall survival (OS) as a hazard ratio (HR) comparing high versus low SII. Risk of bias was assessed with the Newcastle-Ottawa Scale (NOS). Random-effects models were applied to pool HRs, and heterogeneity was summarized with I{superscript 2} and {tau}{superscript 2}. We evaluated publication bias with visual inspection of funnel plots and Eggers regression test. ResultsSeven studies (N= 2,153) met inclusion criteria. High preoperative SII was associated with significantly worse OS (pooled HR 2.17; 95% CI 1.55-2.79). with moderate heterogeneity (I{superscript 2} = 26.0%, {tau}{superscript 2} = 0.0285). Results were robust in leave-one-out analyses, and variability in study-specific SII cut-offs accounted for part of the heterogeneity. Certainty of evidence for the primary outcome was moderate, and all included studies were high quality. Conclusionspreoperative SII is an inexpensive, available biomarker that correlated with risk in resectable GBC and is able to identify patients with more aggressive tumor biology despite Surgical surgery. Systematic review registrationPROSPERO CRD420251185808

ObjectivePostoperative pain outcomes following spinal fusion surgery remain difficult to predict, as structural and surgical indicators alone offer limited insight into who will experience meaningful relief. A substantial proportion of patients continue to report persistent pain after surgery, underscoring the need for objective markers that can help identify those at risk of poor recovery. Peak alpha frequency (PAF) has emerged as a promising trait-like neural signature of pain sensitivity in experimental models, where individuals with slower PAF tend to exhibit heightened pain sensitivity. Yet despite this link, its ability to forecast longer-term postoperative pain trajectories remains unclear. MethodsSeventeen adults undergoing cervical or lumbar fusion surgery were included. Resting-state, eyes-closed EEG was recorded preoperatively and at multiple visits after surgery. PAF was extracted from central electrodes using the centre-of-mass method. Pain intensity was assessed longitudinally on standardised self-report pain scales. Associations between PAF measures and postoperative pain change were examined using correlation analyses, and receiver operating characteristic (ROC) analyses evaluated discrimination of pain responders ([&ge;]50% improvement). ResultsPreoperative peak alpha frequency (PAF) was positively associated with longer-term pain reduction at the 3-month follow-up, but showed no consistent relationship with early postoperative pain. Across pain measures, a consistent pattern emerged across the Brief Pain Inventory (BPI), visual analogue scale (VAS), and numerical rating scale (NRS), but not the verbal rating scale (VRS) or Short-Form McGill (SF-MPQ). At the 3-month follow-up, associations reached statistical significance for BPI-Worst ({rho} = 0.67, p = 0.017), and BPI-Average Pain ({rho} = 0.62, p = 0.033). VAS and NRS showed moderate-to-strong effects that approached significance in non-parametric analyses and were significant for VAS when treated as an approximately interval measure (Pearson r = 0.63, p = 0.022). ROC analyses using BPI-Worst pain improvement demonstrated good discriminative ability of preoperative PAF for identifying treatment responders at 3 months (AUC = 0.84; 95% CI: 0.61-1.00), with high specificity and moderate sensitivity at the Youden-optimal threshold of 10.11 Hz. By contrast, changes in PAF over time were not reliably related to changes in pain scores, suggesting that PAF functions more as a stable, trait-like predictor than a dynamic biomarker in this context. ConclusionThis study demonstrates the feasibility and potential clinical value of preoperative EEG for characterising individual differences in postoperative pain recovery following spinal fusion. The results identify faster preoperative PAF as a stable neural signal that captures meaningful variability in longer-term pain reduction, with convergent support across multiple patient-reported measures. While replication in a larger cohort is required, these findings establish a clear foundation for evaluating PAF as a candidate neurophysiological marker to inform preoperative risk profiling and potentially personalised perioperative pain-management strategies in spinal fusion patients.

AimsObjective identification of patient risk profile in Oesophageal Cancer (OC) surgery is critical. This study aimed to evaluate to what extent cardiorespiratory fitness and select metabolic factors predict clinical outcome. MethodsConsecutive 186 patients were recruited (median age 69 yr. 160 male, 138 neoadjuvant therapy). All underwent pre-operative cardiopulmonary exercise testing to determine peak oxygen uptake [Formula], anaerobic threshold (AT), and ventilatory equivalent for carbon dioxide [Formula]. Cephalic venous blood was assayed for serum C-reactive protein (CRP), albumin, and full blood count. Primary outcome measures were Morbidity Severity Score (MSS), and Overall Survival (OS). ResultsMSS (Clavien-Dindo >2) developed in 33 (17.7%) and was related to elevated CRP (AUC 0.69, p=0.001) and lower V{middle dot}O2Peak (AUC 0.33, p=0.003). Dichotomisation of CRP (above 10mg/L) and V{middle dot}O2Peak (below 18.6mL/kg/min) yielded adjusted Odds Ratios (OR) for MSS CD>2, of 4.01 (p=0.002) and 3.74 (p=0.002) respectively. OC recurrence occurred in 36 (19.4%) and 69 (37.1%) patients died. On multivariable analysis; pTNM stage (Hazard Ratio (HR) 2.20, p=0.001), poor differentiation (HR 2.20, p=0.010), resection margin positivity (HR 2.33, p=0.021), and MSS (HR 4.56, p<0.001) were associated with OS. ConclusionsCRP and V{middle dot}O2Peak are collective independent risk factors that can account for over half of OC survival variance.

BackgroundFumarate hydratase-deficient renal cell carcinoma (FHd-RCC) is a rare and aggressive renal cancer subtype characterised by increased fumarate accumulation and upregulated lactate production. Renal tumours demonstrate significant intratumoral metabolic heterogeneity, which may contribute to treatment failure. Emerging non-invasive metabolic imaging techniques have clinical potential to more accurately phenotype tumour metabolism and its heterogeneity. MethodsHere we have used hyperpolarised 13C-pyruvate MRI (HP 13C-MRI) to assess 13C-lactate generation in a patient with an organ-confined FHd-RCC. Post-operative tissue samples were co-registered with imaging and underwent sequencing, IHC staining, and mass spectrometry imaging (MSI). ResultsHP 13C-MRI revealed two metabolically distinct tumour regions. The 13C-lactate-rich region showed a high lactate/pyruvate ratio and slightly lower fumarate on MSI compared to the other tumour region, as well as increased CD8+ T cell infiltration, and genetic dedifferentiation. Compared to the normal kidney, vascularity in tumour was decreased, while immune cell fraction was markedly higher. ConclusionsThis study shows the potential of metabolic HP 13C-MRI to characterise FHd-RCC and how targeting of biopsies to regions of metabolic dysregulation could be used to obtain the tumour samples of greatest clinical significance, which in turn can inform on early and successful response to treatment.

ObjectivesTo investigate the metabolome perturbation in the progression of hyperuricemia (HUA) into gout, and evaluate the predictive power of metabolomics. MethodsCirculating metabolomics data from 24225 individuals were measured using nuclear magnetic resonance (NMR) technology, and we used the Cox models to assess the hazard ratios of metabolites in HUA-to-gout progression. Key metabolites were selected through 10-fold cross-validated elastic net regression, with 10-year prediction models developed using multivariate Cox regression. The predictive performance was differentiated by comparing the area under the receiver operating characteristic curves (AUCs). We used Net Reclassification Improvement (NRI) to estimate the improvement in reclassification ability with the addition of metabolites to the conventional prediction model. ResultsOf the 24225 HUA patients, the median follow-up period was 13.6 years, during which 1584 participants developed gout. 18 metabolites showed significant associations; the most positive association was with glycoprotein acetyl (HR 1.10; 95% CI: 1.04, 1.16) and the most negative association was with IDL particle concentration (HR 0.91; 95% CI: 0.87, 0.96). The predictive ability (AUC: 0.80 vs 0.78) and reclassification ability (NRI = 2.83%, P < 0.001) of the new combined model were significantly improved with the addition of selected metabolites (n = 44), allowing the identification of high-risk groups. ConclusionsOur analyses identified various metabolomic profiles significantly associated with the development of HUA into an incident event of gout, and implied that metabolomics can enhance predictive accuracy for clinical progression from HUA to gout.

Nonalcoholic fatty liver disease (NAFLD) is the most rapidly growing contributor to chronic liver disease worldwide with high disease burden and suffers from limitations in diagnosis. Inspired by recent advances in machine learning digital diagnostics, we explored the efficacy of training a neural network to classify high risk NAFLD vs. non-NAFLD patients in the UK Biobank dataset based on proton density fat fraction (PDFF). We compared the performance of several ResNet-derived architectures in the context of whole abdomen MRI, segmented liver and abdomen excluding liver (sans-liver). Non-local ResNet trained on whole abdomen MRI images yielded the highest precision (0.88 for NAFLD) and F1 (0.89 for NAFLD). Furthermore, our work on a second, larger cohort explored multi-task learning and the relationship among PDFF, visceral adipose tissue (VAT) and abdominal subcutaneous adipose tissue (ASAT). Interestingly, multi-task learning experiments found a decline in performance for PDFF when combined with VAT and ASAT. We address this deterioration using Multi-gate Mixture-of-Experts (MMoE) approaches. Our work opens the possibility for using a non-invasive deep learning-based diagnostic for NAFLD, and directly enables clinical and genomic research using a larger cohort of potential NAFLD patients in the UK Biobank study.

BackgroundSepsis-associated acute kidney injury (AKI) is a severe condition associated with unfavorable outcomes in critically ill patients, not least because of its delayed diagnosis and management due to limitations in the current standard of care. We aimed to evaluate the performance of sphingotest(R) penKid(R) for the early diagnosis of AKI in patients with sepsis or septic shock. MethodsPlasma proenkephalin A 119-159 (penKid) was measured in 120 healthy subjects and 529 critically ill sepsis patients. Subgroup analyses were performed for patients with and without a history of heart failure or hypertension. The clinical performance for the diagnosis of AKI within 48 hours (AKI48h) was calculated using the upper limit of the penKid reference range. To improve clinical decision-making, interpretation bands and likelihood ratios for optimized rule-in and rule-out performance were established. ResultsOf the 529 patients with sepsis or septic shock, 328 were male (62%), and the median age was 66 (interquartile range 56-75) years. Two hundred thirty-four (44%) patients were diagnosed with AKI48h, and those patients presented increased penKid levels on intensive care unit admission compared with patients without AKI48h, with an area under the curve of 0.87 (95% confidence interval [CI] 0.84-0.90, p<0.0001). A penKid cut-off of 89 pmol/L, corresponding to the 97.5th percentile in the healthy reference population, resulted in 72% sensitivity (95% CI 66-77%), 83% specificity (95% CI 78-87%), 77% positive predictive value (95% CI 71-82%), 79% negative predictive value (95% CI 74-83%), a positive likelihood ratio of 4.2, and a negative likelihood ratio of 0.34 to diagnose AKI48h. An improved performance was obtained when patients with a history of chronic heart failure and/or hypertension were excluded. PenKid cutoff values of 54 pmol/L (>92% sensitivity) and 105 pmol/L (>92% specificity) were derived to establish actionable interpretation bands for diagnostic rule-in and rule-out. ConclusionsThe sphingotest(R) penKid(R) assay can facilitate an early diagnosis of AKI up to 48 hours before the KDIGO criteria are met. Based on the present results, the assay was registered as an aid in the early diagnosis of AKI in patients with sepsis or septic shock.

BACKGROUNDDelayed cerebral ischaemia (DCI) following aneurysmal subarachnoid haemorrhage (aSAH) is a major cause of complications and death. Here we set out to identify high-performance predictive biomarkers of DCI and its underlying metabolic disruptions using metabolomics and lipidomics approaches. METHODSThis single-centre retrospective observational study enrolled 61 consecutive patients with severe aSAH requiring external ventricular drainage between 2013 and 2016. Of these 61 patients, 22 experienced a DCI and were classified as DCI+ and the other 39 patients were classified as DCI-. A further 9 patients with other neurological features were included as non aSAH controls. Blood and cerebrospinal fluid (CSF) were sampled within the first 24 h after admission. We carried out LC-MS/MS-based plasma and CSF metabolomic profiling together with total lipid fatty acids analysis. RESULTSWe identified a panel of 20 metabolites that together showed high predictive performance for DCI (area under the receiver operating characteristic curve: 0.968, specificity: 0.88, sensitivity: 0.94). This panel of metabolites included lactate, cotinine, salicylate, 6 phosphatidylcholines, and 4 sphingomyelins. Analysis of the whole set of metabolites to highlight early biological disruptions that might explain the subsequent DCI found peripheral hypoxia driven mainly by higher blood lactate, arginine and proline metabolism likely associated to vascular NO, dysregulation of the citric acid cycle in the brain, defective peripheral energy metabolism and disrupted ceramide/sphingolipid metabolism. We also unexpectedly found a potential influence of gut microbiota on the onset of DCI. CONCLUSIONWe identified a high-performance predictive metabolomic/lipidomic signature of further DCI in aSAH patients at admission to a NeuroCritical Care Unit. This signature is associated with significant peripheral and cerebral biological dysregulations. We also found evidence, for the first time, pointing to a possible gut microbiota/brain DCI axis, and proposed the putative microorganisms involved. Clinical trial registrationClinicaltrials.gov identifier: NCT02397759

BackgroundMajor Adverse Kidney Events within 30 days (MAKE30) is an important patient-centered outcome for assessing the impact of acute kidney injury (AKI). The existing prediction models for MAKE30 are static and overlook dynamic changes in clinical status. In this study, we introduce ORAKLE, a novel deep-learning model that utilizes evolving time-series data to predict MAKE30, enabling personalized, patient-centered approaches to AKI management and outcome improvement. MethodsWe conducted a retrospective study using three publicly available critical care databases: MIMIC-IV, SICdb, and eICU-CRD. Among these, MIMIC-IV was divided into 80% training and 20% internal test sets, whereas SiCdb and eICU-CRD were used as external validation cohorts. Patients with sepsis-3 criteria who developed AKI within 48 hours of intensive care unit admission were identified. Our primary outcome was MAKE30, defined as a composite of death, new dialysis or persistent kidney dysfunction within 30 days of ICU admission. We developed ORAKLE using Dynamic DeepHit framework for time-series survival analysis and its performance against Cox models using AUROC and AUPRC. We further assessed model calibration using Brier score. ResultsWe analyzed 16,671 patients from MIMIC-IV, 2,665 from SICdb, and 11,447 from eICU-CRD. ORAKLE outperformed the Cox models in predicting MAKE30, achieving AUROCs of 0.84 (95% CI: 0.83-0.86) vs. in MIMIC-IV internal test set 0.80 (95% CI: 0.78-0.82), 0.83 (95% CI: 0.81-0.85) vs. 0.79 (95% CI: 0.77-0.81) in SICdb, and 0.85 (95% CI: 0.84-0.85) vs. 0.81 (95% CI: 0.80-0.82) in eICU-CRD. The AUPRC values for ORAKLE were also significantly better than that of Cox models. The Brier score for ORAKLE was 0.21 across the internal test set, SICdb, and eICU-CRD, suggesting good calibration. ConclusionsORAKLE is a robust deep-learning model for predicting MAKE30 in critically ill patients with AKI that utilizes evolving time series data. By incorporating dynamically changing time series features, the model captures the evolving nature of kidney injury, treatment effects, and patient trajectories more accurately. This innovation facilitates tailored risk assessments and identifies varying treatment responses, laying the groundwork for more personalized and effective management approaches.

ObjectiveTo define phenotypes of gastric myoelectrical abnormalities and relation to symptoms in people with longstanding T1D, compared to matched healthy controls, using a novel non-invasive body surface gastric mapping (BSGM) device. Research design and methodsBSGM was performed on people with T1D of >10 years duration and matched controls, employing Gastric Alimetry (Alimetry, New Zealand), comprising a high-resolution 64-channel array, validated symptom logging App, and wearable reader. Results32 people with T1D were recruited (15 with a high symptom burden), and 32 controls. Those with symptoms showed more unstable gastric myoelectrical activity, (Gastric Alimetry Rhythm Index 0.39 vs 0.51, p=0.017; and lower average spatial covariance 0.48 vs 0.51, p=0.009) compared with controls. Those with T1D and symptoms also had higher prevalence of peripheral neuropathy (67% vs 6%, p=0.001), anxiety/depression diagnoses (27% vs 0%, p=0.001), and mean HbA1c levels (76 vs 56 mmol/mol, p<0.001). BSGM defined distinct phenotypes in participants including those with markedly unstable gastric rhythms (4/32, 12.5%), and abnormally high gastric frequencies (10/32, 31%). Deviation in gastric frequency was positively correlated with symptoms of bloating, upper gut pain, nausea and vomiting, and fullness and early satiation (r>0.35, p<0.05) ConclusionGastroduodenal symptoms in people with longstanding T1D correlate with gastric myoelectrical abnormalities on BSGM evaluation, in addition to glycemic control, psychological comorbidities, and peripheral neuropathy. BSGM using the Gastric Alimetry device identified a range of myoelectrical phenotypes, representing both myogenic and neurogenic mechanisms, which represent targets for diagnosis, monitoring and therapy.