Systematic identification of ACE2 expression modulators reveals cardiomyopathy as a risk factor for mortality in COVID-19 patients

Figures

• 

  Fig. 1. Mining RNA-seq data using Gene Expression Variance Analysis (GENEVA). A The workflow of GENEVA. B The procedure for embedding metadata into numerical vectors

• 

  Fig. 2. Highlighted conditions that modulate ACE2 expression. A,B Plots showing data from a study with GEO accession GSE89714. A A scatter plot showing the association between ACE2 expression and the embedded metadata. All metadata fields are concatenated for embedding. The plot only shows the sample title as labels. The labels are jittered to avoid perfect overlap. The color code represents the ACE2 expression level. B Box plot showing the ACE2 expression in normal hearts and hearts with hypertrophic cardiomyopathy. C,D Plots visualizing data from a study with GEO accession GSE104177, showing the ACE2 expression in breast cancer xenografts with or without RAD140 treatment. E,F Plots visualizing data from a study with GEO accession GSE114013, showing the ACE2 expression in two prostate cancer cell lines with or without itraconazole treatment

• 

  Fig. 3. Meta-analysis confirms ACE2 upregulation in all major types of cardiomyopathy. A The ACE2 expression in normal hearts and hearts with DCM. Data are from five RNA-seq datasets and one microarray dataset. The overall p value is calculated using a mixed model, with the dataset as the random effect and DCM as the fixed effect. B–E The ACE2 expression in normal hearts and hearts with HCM, RCM, LVNC, and ARVC. p values in B–E are calculated using t-tests

• 

  Fig. 4. Patients with pre-existing cardiomyopathy show an increased mortality rate. A Kaplan-Meier curve of COVID19 patients with pre-existing cardiomyopathy (N = 43), patients with other pre-existing cardiovascular diseases (N = 624), and patients without cardiovascular diseases (N = 3269). P values are from Cox proportional-hazards models, controlling for differences in the demographics and non-cardiovascular conditions between the groups. B Kaplan-Meier curve of COVID19 patients with pre-existing cardiomyopathy (N = 43) and a cohort of propensity score-matched patients with other pre-existing cardiovascular diseases (N = 344). P values are from a log-rank test. C Kaplan-Meier curve of COVID-19-negative patients with pre-existing cardiomyopathy (N = 2250) and a cohort of propensity score-matched patients with other pre-existing cardiovascular diseases (N = 18,000). P values are from a log-rank test