Modeling of senescence-related chemoresistance in ovarian cancer using data analysis and patient-derived organoids

Abstract

Background: Ovarian cancer (OC) is a malignant tumor associated with poor prognosis owing to its susceptibility to chemoresistance. Cellular senescence, an irreversible biological state, is intricately linked to chemoresistance in cancer treatment. We developed a senescence-related gene signature for prognostic prediction and evaluated personalized treatment in patients with OC.

Methods: We acquired the clinical and RNA-seq data of OC patients from The Cancer Genome Atlas and identified a senescence-related prognostic gene set through differential and cox regression analysis in distinct chemotherapy response groups. A prognostic senescence-related signature was developed and validated by OC patient-derived-organoids (PDOs). We leveraged gene set enrichment analysis (GSEA) and ESTIMATE to unravel the potential functions and immune landscape of the model. Moreover, we explored the correlation between risk scores and potential chemotherapeutic agents. After confirming the congruence between organoids and tumor tissues through immunohistochemistry, we measured the IC₅₀ of cisplatin in PDOs using the ATP activity assay, categorized by resistance and sensitivity to the drug. We also investigated the expression patterns of model genes across different groups.

Results: We got 2740 differentially expressed genes between two chemotherapy response groups including 43 senescence-related genes. Model prognostic genes were yielded through univariate cox analysis, and multifactorial cox analysis. Our work culminated in a senescence-related prognostic model based on the expression of SGK1 and VEGFA. Simultaneously, we successfully constructed and propagated three OC PDOs for drug screening. PCR and WB from PDOs affirmed consistent expression trends as those of our model genes derived from comprehensive data analysis. Specifically, SGK1 exhibited heightened expression in cisplatin-resistant OC organoids, while VEGFA manifested elevated expression in the sensitive group (P<0.05). Intriguingly, GSEA results unveiled the enrichment of model genes in the PPAR signaling pathway, pivotal regulator in chemoresistance and tumorigenesis. This revelation prompted the identification of potential beneficial drugs for patients with a high-risk score, including gemcitabine, dabrafenib, epirubicin, oxaliplatin, olaparib, teniposide, ribociclib, topotecan, venetoclax.

Conclusion: Through the formulation of a senescence-related signature comprising SGK1 and VEGFA, we established a promising tool for prognosticating chemotherapy reactions, predicting outcomes, and steering therapeutic strategies. Patients with high VEGFA and low SGK1 expression levels exhibit heightened sensitivity to chemotherapy.

Keywords: TCGA; cell senescence; chemoresistance; organoid; ovarian cancer.

Figure 1

The flowchart for constructing and passaging ovarian cancer organoids.

Figure 2

Identification of 12 vital differentially expressed senescence genes in ovarian cancer. Heatmap (A) and volcano plot (C) of DEGs between the platinum-resistant and sensitive OC patients based on TCGA. (B) Venn diagram of ageing-genes and DEGs, (D) Univariate Cox analysis of the aging-related DEGs expression. (E) PPI network illustrated the relationship among the 12 DEGs. (F) Hallmark enrichment analysis of the aging-related DEGs. (*P < 0.05).

Figure 3

Construction and verification of the prognostic index. (A) Forest plot of aging-related DEGs via multivariate Cox regression analysis. (B) The survival ability of OC patients with high and low expression level of VEGFA (P=0.0168) and SGK1 (P=0.0137). Q-Q plot of VEGFA and SGK1 to test the normal distribution of data. (C) The expression of VEGFA and SGK1 in resistant and sensitive groups. (D) Differences in overall survival between high- and low-risk groups (P=0.0198). (E) Mutation waterfall maps show the gene mutation differences in high- and low-risk groups. (* P<0.05).

Figure 4

Tumor immune microenvironment. (A) t-SNE plot visualized 9 cell subtypes in OC patients. (B) The annotation diagram of different cell types. (C, D) The distribution of SGK1 and VEGFA expression in all cell types.

Figure 5

GSEA of aging-related DEGs based on hallmark gene sets (A) and KEGG database (B). (NES: normalized enriched score).

Figure 6

Different proportion of 22 immune cells between high- and low-risk groups analyzed by CIBERSORT algorithm (A) and heatmap (C). (B) Estimate analysis of immune, stromal, and tumor purity score in high- and low-risk group.

Figure 7

Correlation between risk score and IC₅₀ of potential chemotherapeutics predicted by the “pRRophetic” package.

Figure 8

(A) The growth states of organoids recorded. (Scale bar = 100 µm) (The red arrow indicates the same organoid) (B) HE and IHC staining of P53, MUC16, WT-1 in OC tissues and organoids (Scale bar for tissue = 70 µm, scale bar for organoid = 40 µm).

Figure 9

(A) The different state of organoids in two groups after culturing 72 hours with 12.5 μM cisplatin. The red arrows indicate organoid changes. (B) The inhibition ratio plot of organoids under different concentrations cisplatin. Data presented as mean ± SD. (C) Relative expression level of SGK1 and VEGFA in organoids. Data presented as the mean ± SEM, one-way ANOVA was used for statistical calculation, n=3 independent experiments. (D) Western blot and gray values of SGK1 and VEGFA expression in PDOs. (Scale bar = 100 µm, ns: No significance **P < 0.01. ***P < 0.001, ****P < 0.0001).