{
  "abstract": "Objectives To predict existing nephritis and potential nephritic tendencies in Lupus patients by building a machine learning model using their available electronic case report form (eCRF) data and combining it with a liquid biopsy strategy analysing by performing miRNASeq analysis from urinary extracellular vesicles (uEVs).Methods eCRF data of 390 patients from the PRECISESADS cohort were used to train and build a machine learning model that was validated on a held-out test set of 74 patients. These 74 patients were then regrouped by comparing their classification done by the model and the ground truth, and small RNAseq was performed on their uEVs, along with that of 30 healthy donors. Data was analysed keeping batch and other probable covariates in the design to find out dysregulated miRNAs for each group of patients. Following this, meta-analysis was performed to find out shared effects of miRNAs among different groups and genes and pathways related to them were investigated. Further comparison with patients of other autoimmune disorders were also made to find out shared effect of these miRNAs. An optimized method for uEV isolation from small volume of urine was developed to validate the miRNAs with digital PCR, which is being done in the present.Results According to the predictive model results, patients were classified into ‘No Nephritis’ (N = 33), ‘Predicted Nephritis’ (N = 18) and ‘Established Nephritis’ (N = 23) groups. miRNAs hsa-miR-9-5p, hsa-miR-181a-2-3p, hsa-miR132-3p are markedly dysregulated in the ‘Predicted Nephritis’ group. Moreover, miRNAs hsa-miR-146a-5p, hsa-miR-192-5p are dysregulated in the ‘Established Nephritis’ group, and miRNAs hsa-miR-335-3p, hsa-miR-455-3p, hsa-miR-143-3p are dysregulated across different nephritis stages. Finaly, miRNAs hsa-miR-126-5p, hsa-miR-888-5p and hsa-miR-501-5p were significant only in nephritis specific groups, and they were associated with genes such as INO80D, MAFB, QKI and ferroptosis and calcium pathways, among others.Abstract PO:02:047 Figure 1Conclusions Our model contributes to classify patients according to its nephritis status by combining eCRF data with a panel of uEV miRNAs. This liquid biopsy strategy can help clinicians by guiding therapeutic related decisions to manage LN in a more personalized manner. The selected miRNA pannel is being validated in uEVs from small volume of urines using digital PCR.",
  "authors": [
    {
      "affiliations": [
        "Atrys Health SA, Barcelona, Spain"
      ],
      "name": "Hindol Mazumdar"
    },
    {
      "affiliations": [
        "Atrys Health SA, Barcelona, Spain"
      ],
      "name": "Judith Gonzalez Garcia"
    },
    {
      "affiliations": [
        "Centro Pfizer-Universidad de Granada-Junta de Andalucía de Genómica e Investigación Oncológica (GENYO), Granada, Spain",
        "University of Granada, Department of Genetics, Faculty of Science, Granada, Spain",
        "Bioinformatics Laboratory, Biotechnology Institute, Centro de Investigación Biomédica, PTS, Granada, Spain"
      ],
      "name": "Guillermo Barturen"
    },
    {
      "affiliations": [
        "Centro Pfizer-Universidad de Granada-Junta de Andalucía de Genómica e Investigación Oncológica (GENYO), Granada, Spain"
      ],
      "name": "PRECISESADS Clinical Consortium"
    },
    {
      "affiliations": [
        "Atrys Health SA, Barcelona, Spain"
      ],
      "name": "Victor Gonzalez Rumayor"
    },
    {
      "affiliations": [
        "Centro Pfizer-Universidad de Granada-Junta de Andalucía de Genómica e Investigación Oncológica (GENYO), Granada, Spain",
        "Karolinska Institute – Institute for Environmental Medicine, Stockholm, Sweden"
      ],
      "name": "Marta Eugenia Alarcon Riquelme"
    },
    {
      "affiliations": [
        "Atrys Health SA, Barcelona, Spain"
      ],
      "name": "Nadina Erill Sagales"
    },
    {
      "affiliations": [
        "Centro Pfizer-Universidad de Granada-Junta de Andalucía de Genómica e Investigación Oncológica (GENYO), Granada, Spain"
      ],
      "name": "Concepción Marañón"
    }
  ],
  "title": "PO:02:047 Using machine learning and liquid biopsy strategies to predict potential silent nephritis in systemic lupus erythematosus patients",
  "uid": "91c3af2c-f3f4-5091-b6c0-748dd46ca8e0"
}