Longitudinal peripheral blood transcriptional analysis of COVID-19 patients captures disease progression and reveals potential biomarkers

Qihong Yan, Pingchao Li, Xianmiao Ye, Xiaohan Huang, Xiaoneng Mo, Qian Wang, Yudi Zhang, Kun Luo, Zhaoming Chen, Jia Luo, Xuefeng Niu, Ying Feng, Tianxing Ji, Bo Feng, Jinlin Wang, Feng Li, Fuchun Zhang, Fang Li, Jianhua Wang, Liqiang Feng, Zhilong Chen, Chunliang Lei, Linbing Qu, Ling ChenmedRxiv 2020 · DOI

Abstract

COVID-19, caused by SARS-CoV-2, is an acute self-resolving disease in most patients, but some can develop severe illness or even death. To characterize host responses and identify potential biomarkers during disease progression, we performed a longitudinal transcriptome analysis for peripheral blood mononuclear cells (PBMCs) collected from four COVID-19 patients at four different time points from symptom onset to recovery. We found that PBMCs at different COVID-19 disease stages exhibited unique transcriptome characteristics. SARS-CoV-2 infection dysregulated innate immunity, especially the type I interferon response, and disturbed the release of inflammatory cytokines and lipid mediators. An aberrant increase of low-density neutrophils (LDNs) may cause tissue damage, while the activation of cell death, exhaustion, and migratory pathways may lead to the reduction of lymphocytes and dysfunction of adaptive immunity. Furthermore, COVID-19-induced hypoxia may exacerbate disorders in blood coagulation. Based on our analysis, we proposed a set of potential biomarkers for monitoring disease progression and predicting the risk of severity.

Introduction

One of the most challenging aspects of the COVID-19 pandemic has been the clinical unpredictability of the virus. While many patients experience mild or self-limiting symptoms, a subset progresses rapidly to severe respiratory failure and systemic complications. Traditional clinical markers, such as C-reactive protein (CRP), often lack the sensitivity to predict this transition before a patient reaches a critical state.

This study sought to map the host immune response at a high resolution by tracking the "transcriptional fingerprints" of blood cells over time. By analyzing how gene expression evolves from the first signs of symptoms through to recovery, the researchers aimed to identify molecular "tripwires"—biomarkers that could signal a high risk of severe disease earlier than traditional clinical observations.

Methodology

The researchers employed longitudinal transcriptome analysis, sequencing the messenger RNA (mRNA) of Peripheral Blood Mononuclear Cells (PBMCs) at four distinct clinical stages:

  1. Stage 1: Early onset of symptoms.
  2. Stage 2: Peak of disease severity.
  3. Stage 3: Clinical improvement.
  4. Stage R: Recovery.

To analyze this data, the team used Digital Cytometry (CIBERSORTx) to estimate the abundance of specific immune cell types and Time-series Clustering (Mfuzz) to group genes with similar expression trajectories. Finally, a Multi-category Classification (MCC) approach was applied to identify the most predictive biomarkers for disease stages and severity. The primary cohort consisted of 16 samples from four female patients, validated against three additional severe cases and healthy controls.

Key Biological Insights

The "Silent" Interferon Gap

A critical finding was the dysregulation of the innate immune response. While Interferon-Stimulated Genes (ISGs)—the body's primary antiviral defense—were highly active during the early stages, the actual Type I Interferon transcripts (IFN-α/β) were nearly undetectable. This suggests a paradox where the virus may actively suppress the production of these signaling proteins, even as the downstream "alarm system" remains primed.

Low-Density Neutrophils and Tissue Damage

The study observed a dramatic increase in low-density neutrophils (LDNs) and the expression of genes related to Neutrophil Extracellular Traps (NETs). While NETs are designed to trap pathogens, their excessive and unregulated release in COVID-19 patients appears to cause significant collateral tissue damage and promote the formation of micro-clots.

The Hypoxia-Coagulation Axis

The analysis provided a molecular explanation for the blood clotting issues frequently seen in severe COVID-19. The researchers found that infection-induced hypoxia (low oxygen) triggers genes like HIF1A, which in turn upregulates pro-coagulant factors. This link between respiratory distress and disseminated intravascular microthrombosis highlights a dangerous feedback loop in severe cases.

Lipid Mediator Dysregulation

The study also identified a surge in pro-inflammatory lipid mediators, such as PGE2, which may inhibit the antiviral response. Conversely, a transient spike in the anti-inflammatory mediator LXA4 during early stages suggests the body's initial, though often insufficient, attempt to self-regulate the inflammatory storm.

Identification of a 12-Gene Biomarker Signature

By applying machine learning to the longitudinal data, the researchers identified a specific 12-gene signature that accurately mirrors disease progression and predicts severity. The panel includes: IL6, IL10, CXCL2, ALOX15B, IL1R2, MMP9, GRP84, F3, SERPINE1, CD8B, CD3G, and S1PR1.

In the validation cohort, this 12-gene combination achieved an Area Under the Curve (AUC) of 1.0, perfectly distinguishing patients who would later require ICU care from those who would not, based on samples taken during the early onset stage. This suggests that these genes could serve as a powerful early warning system for clinicians.

Conclusion and Future Directions

This longitudinal map provides a high-resolution view of the immune system's battle with SARS-CoV-2. While the study is limited by a small, single-gender primary cohort (four female patients), the high predictive power of the 12-gene signature warrants further investigation in larger, more diverse populations. Future work using single-cell RNA sequencing (scRNA-seq) could further refine these findings by identifying the specific cell subsets driving these transcriptional changes. Ultimately, translating these 12 biomarkers into a rapid, hospital-ready diagnostic test could significantly improve the triage and treatment of high-risk COVID-19 patients.

Summary updated

Abstract

ABSTRACT COVID-19, caused by SARS-CoV-2, is an acute self-resolving disease in most of the patients, but some patients can develop a severe illness or even death. To characterize the host responses and identify potential biomarkers during disease progression, we performed a longitudinal transcriptome analysis for peripheral blood mononuclear cells (PBMCs) collected from 4 COVID-19 patients at 4 different time points from symptom onset to recovery. We found that PBMCs at different COVID-19 disease stages exhibited unique transcriptome characteristics. SARS-CoV-2 infection dysregulated innate immunity especially type I interferon response as well as the disturbed release of inflammatory cytokines and lipid mediators, and an aberrant increase of low-density neutrophils may cause tissue damage. Activation of cell death, exhaustion and migratory pathways may lead to the reduction of lymphocytes and dysfunction of adaptive immunity. COVID-19 induced hypoxia may exacerbate disorders in blood coagulation. Based on our analysis, we proposed a set of potential biomarkers for monitoring disease progression and predicting the risk of severity.
BibTeX
@misc{yan2020longitudinal,
  author = {Qihong Yan and Pingchao Li and Xianmiao Ye and Xiaohan Huang and Xiaoneng Mo and Qian Wang and Yudi Zhang and Kun Luo and Zhaoming Chen and Jia Luo and Xuefeng Niu and Ying Feng and Tianxing Ji and Bo Feng and Jinlin Wang and Feng Li and Fuchun Zhang and Fang Li and Jianhua Wang and Liqiang Feng and Zhilong Chen and Chunliang Lei and Linbing Qu and Ling Chen and Linbing Qu and Ling Chen},
  title = {{Longitudinal peripheral blood transcriptional analysis of COVID-19 patients captures disease progression and reveals potential biomarkers}},
  year = {2020},
  doi = {10.1101/2020.05.05.20091355},
  howpublished = {medRxiv},
}
Figure