Predictive models for delayed graft function (DGF) after kidney transplantation are usually developed using logistic regression. We want to evaluate the value of machine learning methods in the prediction of DGF. 497 kidney transplantations from deceased donors at the Ghent University Hospital between 2005 and 2011 are included. A feature elimination procedure is applied to determine the optimal number of features, resulting in 20 selected parameters (24 parameters after conversion to indicator parameters) out of 55 retrospectively collected parameters. Subsequently, 9 distinct types of predictive models are fitted using the reduced data set: logistic regression (LR), linear discriminant analysis (LDA), quadratic discriminant analysis (QDA), support vector machines (SVMs; using linear, radial basis function and polynomial kernels), decision tree (DT), random forest (RF), and stochastic gradient boosting (SGB). Performance of the models is assessed by computing sensitivity, positive predictive values and area under the receiver operating characteristic curve (AUROC) after 10-fold stratified cross-validation. AUROCs of the models are pairwise compared using Wilcoxon signed-rank test. The observed incidence of DGF is 12.5 %. DT is not able to discriminate between recipients with and without DGF (AUROC of 52.5 %) and is inferior to the other methods. SGB, RF and polynomial SVM are mainly able to identify recipients without DGF (AUROC of 77.2, 73.9 and 79.8 %, respectively) and only outperform DT. LDA, QDA, radial SVM and LR also have the ability to identify recipients with DGF, resulting in higher discriminative capacity (AUROC of 82.2, 79.6, 83.3 and 81.7 %, respectively), which outperforms DT and RF. Linear SVM has the highest discriminative capacity (AUROC of 84.3 %), outperforming each method, except for radial SVM, polynomial SVM and LDA. However, it is the only method superior to LR. The discriminative capacities of LDA, linear SVM, radial SVM and LR are the only ones above 80 %. None of the pairwise AUROC comparisons between these models is statistically significant, except linear SVM outperforming LR. Additionally, the sensitivity of linear SVM to identify recipients with DGF is amongst the three highest of all models. Due to both reasons, the authors believe that linear SVM is most appropriate to predict DGF.

Diffuse large B-cell lymphoma (DLBCL) is the most common histological subtype of non-Hodgkin’s lymphomas (NHL). DLBCL is an aggressive malignancy that displays a great heterogeneity in terms of morphology, genetics and biological behavior. While a sustained complete remission is obtained in the majority of patients with standard immunochemotherapy, patients with refractory of relapsed disease after first-line treatment have a poor prognosis. This patient group represents an important unmet need in lymphoma treatment. In recent years, improved understanding of the underlying molecular pathogenesis had led to new classification and prognostication tools, including the development of cell-free biomarkers in liquid biopsies. Although the majority of studies have focused on the use of cell-free fragments of DNA (cfDNA), there has been an increased interest in circulating-free coding and non-coding RNA, including messenger RNA (mRNA), microRNA (miRNA), long non-coding RNA (lncRNA) and circular RNA (circRNA), as well as RNA encapsulated in extracellular vesicles or tumor-educated platelets (TEPs). We performed a systematic search in PubMed to identify articles that evaluated circulating RNA as diagnostic, subtype, treatment response or prognostic biomarkers in a human DLBCL population. A total of 35 articles met the inclusion criteria. The aim of this systematic review is to present the current understanding of circulating RNA molecules as biomarker in DLBCL and to discuss their future potential.

Background: This study compared 22 different definitions of delayed graft function (DGF) following kidney transplantation. Material/Methods: Our study included 497 kidney transplantations from deceased donors at our center between 2005 and 2011. Graft survival analysis including log-rank tests and Cox proportional hazards model was performed. Sensitivity and specificity were calculated in relation to graft failure. Results: Mean follow-up time was 5.1 years. All dialysis-based definitions were associated with graft failure and characterized by high specificity (88-97%), but low sensitivity (25-29%). Hazard ratios ranged from 2.87 to 13.73, with increased risk when dialysis was required earlier and more frequently. The urine output-based definition performed similarly, with an association with graft failure and high specificity (96%), but low sensitivity (21%). Serum creatinine-based definitions were more heterogeneous. Higher sensitivity (4-67%) was found in some of these definitions, but was often associated with lower specificity (47-96%), losing the association with graft failure. Definitions combining different criteria varied in sensitivity (17-62%) and specificity (60-96%). However, some were able to achieve higher sensitivity without compromising too much on specificity, while keeping the association with graft failure. Conclusions: Our results indicate a potential advantage of combined definitions, because they are able to detect a larger group of recipients with increased risk of graft failure.

Archived formalin-fixed, paraffin-embedded (FFPE) tissue samples are being increasingly used in molecular cancer research. Compared with fresh-frozen tissue, the nucleic acid analysis of FFPE tissue is technically more challenging. This study aimed to compare the impact of 3 different RNA extraction methods on yield, quality, and sequencing-based gene expression results in FFPE samples. RNA extraction was performed in 16 FFPE tumor specimens from patients with diffuse large B-cell lymphoma and in reference FFPE material from microsatellite-stable and microsatellite-instable cell lines (3 replicates each) using 2 silica-based procedures (A, miRNeasy FFPE; C, iCatcher FFPE Tissue RNA) and 1 isotachophoresis-based procedure (B, Ionic FFPE to Pure RNA). The RNA yield; RNA integrity, as reflected by the distribution value 200; and RNA purity, as reflected by the 260/280 and the 260/230 nm absorbance ratios, were determined. The RNA was sequenced on the NovaSeq 6000 instrument using the TruSeq RNA Exome and SMARTer Stranded Total RNA-Seq Pico v3 library preparations kits. Our results highlight the impact of RNA extraction methodology on both preanalytical and sequencing-based gene expression results. Overall, methods B and C outperformed method A because these showed significantly higher fractions of uniquely mapped reads, an increased number of detectable genes, a lower fraction of duplicated reads, and better representation of the B-cell receptor repertoire. Differences among the extraction methods were generally more explicit for the total RNA sequencing method than for the exome-capture sequencing method. Importantly, the predicative value of quality metrics varies among extraction kits, and caution should be applied when comparing and interpreting results obtained using different methods.

Introduction: Diffuse large B-cell lymphoma (DLBCL) and primary mediastinal B-cell lymphoma (PMBCL) are aggressive histological subtypes of non-Hodgkin’s lymphoma. Improved understanding of the underlying molecular pathogenesis has led to new classification and risk stratification tools, including the development of cell-free biomarkers through liquid biopsies. The goal of this study was to investigate cell-free RNA (cfRNA) biomarkers in DLBCL and PMBCL patients.Materials and methods: Blood plasma samples (n=168) and matched diagnostic formalin-fixed paraffin-embedded (FFPE) tissue samples (n=69) of DLBCL patients, PMBCL patients and healthy controls were collected between 2016-2021. Plasma samples were collected at diagnosis, at interim evaluation, after treatment, and in case of refractory or relapsed disease. RNA was extracted from 200 µl plasma using the miRNeasy serum/plasma kit and from FFPE tissue using the miRNeasy FFPE kit. RNA was subsequently sequenced on a NovaSeq 6000 instrument using the SMARTer Stranded Total RNA-seq pico v3 library preparation kit.Results: Higher cfRNA concentrations were demonstrated in lymphoma patients compared to healthy controls. A large number of differentially abundant genes were identified between the cell-free transcriptomes of DLBCL patients, PMBCL patients, and healthy controls. Overlap analyses with matched FFPE samples showed that blood plasma has a unique transcriptomic profile that significantly differs from that of the tumor tissue. As a good concordance between tissue-derived gene expression and the immunohistochemistry Hans algorithm for cell-of-origin (COO) classification was demonstrated in the FFPE samples, but not in the plasma samples, a 64-gene cfRNA classifier was developed that can accurately determine COO in plasma. High plasma levels of a 9-gene signature (BECN1, PRKCB, COPA, TSC22D3, MAP2K3, UQCRHL, PTMAP4, EHD1, NAP1L1 pseudogene) and a 5-gene signature (FTH1P7, PTMAP4, ATF4, FTH1P8, ARMC7) were significantly associated with inferior progression-free and overall survival in DLBCL patients, respectively, independent of the NCCN-IPI score.Conclusion: Total RNA sequencing of blood plasma samples allows the analysis of the cell-free transcriptome in DLBCL and PMBCL patients and demonstrates its unexplored potential in identifying diagnostic, cell-of-origin, and prognostic cfRNA biomarkers.

Extracellular RNA (cell‐free RNA; exRNA) from blood‐derived liquid biopsies is an appealing, minimally invasive source of disease biomarkers. As pre‐analytical variables strongly influence exRNA measurements, their reporting is essential for meaningful interpretation and replication of results. The aim of this review was to chart to what extent pre‐analytical variables are documented, to pinpoint shortcomings and to improve future reporting. In total, 200 blood plasma exRNA studies published in 2018 or 2023 were reviewed for annotation of 22 variables associated with blood collection, plasma preparation, and RNA purification. Our results show that pre‐analytical variables are poorly documented, with only three out of 22 variables described in over half of the publications. The percentage of variables reported ranged from 4.6% to 54.6% (mean 24.84%) in 2023 and from 4.6% to 57.1% (mean 28.60%) in 2018. Recommendations and guidelines (i.e., BRISQ, ASCO‐CAP, BloodPAC, PPMPT, and CEN standards) have currently not resulted in improved reporting. In conclusion, our results highlight the lack of reporting pre‐analytical variables in exRNA studies and advocate for a consistent use of available standards, endorsed by funders and journals.