Rethinking Lupus Nephritis Classification on a Molecular Level.
Salem AlmaaniStephenie D ProkopecJianying ZhangLianbo YuCarmen Avila-CasadoJoan WitherJames W ScholeyValeria AlbertonAna MalvarSamir V ParikhPaul C BoutrosBrad H RovinHeather N ReichPublished in: Journal of clinical medicine (2019)
The International Society of Nephrology/Renal Pathology Society (ISN/RPS) lupus nephritis (LN) classification is under reconsideration, given challenges with inter-rater reliability and resultant inconsistent relationship with treatment response. Integration of molecular classifiers into histologic evaluation can improve diagnostic precision and identify therapeutic targets. This study described the relationship between histological and molecular phenotypes and clinical responses in LN. Renal compartmental mRNA abundance was measured in 54 biopsy specimens from LN patients and correlated to ISN/RPS classification and individual histologic lesions. A subset of transcripts was also evaluated in sequential biopsies of a separate longitudinal cohort of 36 patients with paired samples obtained at the time of flare and at follow up. Unsupervised clustering based on mRNA abundance did not demonstrate a relationship with the (ISN/RPS) classification, nor did univariate statistical analysis. Exploratory analyses suggested a correlation with individual histologic lesions. Glomerular FN1 (fibronectin), SPP1 (secreted phosphoprotein 1), and LGALS3 (galectin 3) abundance correlated with disease activity and changed following treatment. Exploratory analyses suggested relationships between specific transcripts and individual histologic lesions, with the important representation of interferon-regulated genes. Our findings suggested that the current LN classification could be refined by the inclusion of molecular descriptors. Combining molecular and pathologic kidney biopsy phenotypes may hold promise to better classify disease and identify actionable treatment targets and merits further exploration in larger cohorts.
Keyphrases
- machine learning
- deep learning
- disease activity
- systemic lupus erythematosus
- ultrasound guided
- end stage renal disease
- big data
- transcription factor
- rheumatoid arthritis patients
- lymph node
- prognostic factors
- gene expression
- fine needle aspiration
- neoadjuvant chemotherapy
- artificial intelligence
- chronic kidney disease
- juvenile idiopathic arthritis
- cross sectional
- smoking cessation
- genome wide identification
- bioinformatics analysis