317 - Exploring Single celled RNA Biomarkers using deconvolution analysis for severe Retinopathy of Prematurity in extremely premature newborns

Publication Number: 4311.317

Vighnesh Ghatpande, University of Texas at Austin, Austin, TX, United States; Saachi A. Bodas, Wayne State University School of Medicine, Novi, MI, United States; Nitin Chouthai, Oakland University William Beaumont School of Medicine, Dearborn, MI, United States

Background: Timely, precise diagnosis of Retinopathy of Prematurity (ROP) is key to implementing appropriate treatment, as distinctions between mild-ROP and severe-ROP are the primary factors in its characterization. Deconvolution analysis of bulk RNA sequencing is an indirect method to identify static and dynamic molecular and physiological associations.

Objective: The objective of this this study was to evaluate if deconvolution analysis can be used to identify cellular biomarkers of ROP.

Design/Methods: This is a prospective study (IRB#086113MP2E) with a convenience sample size of 8 neonates with no ROP, 6 neonates with mild to moderate ROP (M-ROP), and 7 neonates with severe ROP (S-ROP) that necessitated laser surgery born at less than 26 weeks gestation. Peripheral blood Mononuclear cells were used for total RNA isolation. The isolated RNA was submitted to the Genome Sciences Core at Wayne State
University for RNA sequencing (RNA-seq). The barcoded libraries were multiplexed at equimolar
concentrations and sequenced (75 bp reads; >5M reads per sample) on the NovaSeq 6000. Data was
demultiplexed using Illumina's CASAVA 1.8.2 software.
In order to identify blood cell proportions in the bulk RNASeq data, a deconvolution analysis was performed using xCell2 1.3.0 R package (https://bioconductor.org/packages/devel/bioc/html/xCell2.html). The pretrained Immune Compendium reference was used to deconvolute the data and identify relative proportions of 40 different blood cell types. To identify significantly different cell populations across the three groups, a pairwise Students t-test was performed and adjusted using an FDR of 0.05.

Results: The demographic characteristics were not statistically significantly different (Table1). The pairwise comparison results are presented in Table 2 and graphically in Figure 1. Amongst 40 cell types evaluated only Neutrophils were statistically significantly higher in newborns Severe ROP as compared to controls. There was no statistically significant difference when data was compared for mild/moderate ROP versus the control group. There was no difference in expression of hematopoietic and endothelial cell types when compared using pan cancer single cell RNA deconvolution analysis (data not shown).

Conclusion(s): This is a unique preliminary data with interesting results. Neutrophils were upregulated in extremely premature newborns at that subsequently developed severe ROP while there were no differences in hemopoietic and endothelial cell types. A prospective single cell RNA analysis is warranted to further identify mechanisms associated with development of severe ROP.

Table .1: Demographic Parameters of Neonates
P values represent out come of One Way ANOVA analysis for continuous variables and Chi Square or Fischer's Exact test for categorical variables.

Table 2: Cell types and adjusted p values in presence of severe ROP (S), Moderate ROP (M) and No Rop (Contol (c ))
P adj values represent the P values adjusted for multiple comparisons.

Figure 1: Bar chart of comparison of Cell types between No ROP, Mild ROP and Severe ROP
Neutrophils were upregulated in presence of severe ROP in extremely premature newborns between day 7 and 14 of life.