Skip to main content
Figure 1 | Genome Biology

Figure 1

From: Combined image and genomic analysis of high-grade serous ovarian cancer reveals PTEN loss as a common driver event and prognostic classifier

Figure 1

PTEN expression in TCGA samples correlates with ACTA2 expression, and thus stromal content. (A) Example of H&E stained sections from TCGA samples having low and high stromal content. The stromal content detected using the segmentation algorithm is shown in green. (B) Average expression of combined stromal signature correlated well with automated quantification in (A) (r=0.397, N=216 patients). (C) Univariate correlation testing of stromal genes from Yoshihara with image analysis showing that ACTA2 is highly correlated with stromal content (r=0.306) [22]. (D) Scatter plot showing the distribution of PTEN vs ACTA2 mRNA expression for each sample (Pearson correlation, r=0.152; P<0.001) with corresponding box plots showing distribution of PTEN expression in different ACTA2 expression quartiles. There is higher PTEN expression with high ACTA2 expression (Jonchkeere Terpsa test, P=0.001). (E) Heat maps of top 50 differentially expressed genes between PTENhigh and PTENlow tumours in the TCGA data set before and after selecting for ACTA2 low samples. In each case, the top quartile of PTEN expression was labelled PTENhigh, and the lowest quartile PTENlow. Correction for ACTA2 content reveals AR as one of the top differentially expressed genes. (F) Stromal gene set enrichment plots after differential expression analysis between high and low PTEN. Stromal-related genes from the Yoshihara signature (141 genes, highlighted in red) are redistributed [22]. There is less enrichment for stromal-related genes after correcting for stroma content (enrichment score 0.5 to 0.1). Dotted lines indicate adjusted P=0.05.

Back to article page