Neoadjuvant Chemotherapy Induces Genomic and Transcriptomic Changes in Ovarian Cancer

Abstract

The growing use of neoadjuvant chemotherapy to treat advanced stage high-grade serous ovarian cancer (HGSOC) creates an opportunity to better understand chemotherapy-induced mutational and gene expression changes. Here we performed a cohort study including 34 patients with advanced stage IIIC or IV HGSOC to assess changes in the tumor genome and transcriptome in women receiving neoadjuvant chemotherapy. RNA sequencing and panel DNA sequencing of 596 cancer-related genes was performed on paired formalin-fixed paraffin-embedded specimens collected before and after chemotherapy, and differentially expressed genes (DEG) and copy-number variations (CNV) in pre- and post-chemotherapy samples were identified. Following tissue and sequencing quality control, the final patient cohort consisted of 32 paired DNA and 20 paired RNA samples. Genomic analysis of paired samples did not reveal any recurrent chemotherapy-induced mutations. Gene expression analyses found that most DEGs were upregulated by chemotherapy, primarily in the chemotherapy-resistant specimens. AP-1 transcription factor family genes (FOS, FOSB, FRA-1) were particularly upregulated in chemotherapy-resistant samples. CNV analysis identified recurrent 11q23.1 amplification, which encompasses SIK2. In vitro, combined treatment with AP-1 or SIK2 inhibitors with carboplatin or paclitaxel demonstrated synergistic effects. These data suggest that AP-1 activity and SIK2 copy-number amplification are induced by chemotherapy and may represent mechanisms by which chemotherapy resistance evolves in HGSOC. AP-1 and SIK2 are druggable targets with available small molecule inhibitors and represent potential targets to circumvent chemotherapy resistance. SIGNIFICANCE: Genomic and transcriptomic analyses identify increased AP-1 activity and SIK2 copy-number amplifications in resistant ovarian cancer following neoadjuvant chemotherapy, uncovering synergistic effects of AP-1 and SIK2 inhibitors with chemotherapy.

Fig. 1.. Transcriptomic and genomic characterization of HGSOC response to neoadjuvant chemotherapy.

(A) Overview of cohort and sample collection time points. Representative H&E of matched pairs of pre- and post-NACT samples from tumors with poor pathologic response to therapy (CRS 1) or excellent response to therapy (CRS 3; 40× magnification). (B) RNA-sequencing of whole transcriptome showed that samples cluster based on treatment status (pre- or post-NACT) rather than chemo-sensitivity, as visualized with t-SNE. CRS = chemotherapy response score; NACT = neoadjuvant chemotherapy; t-SNE = t-distributed stochastic neighbor embedding. (C) Molecular subtype of indicated patients before and after NACT. Patient molecular subtypes are plastic and frequently change after NACT.

Fig. 2.. Neoadjuvant chemotherapy is associated with increased AP-1 activity.

(A) Hierarchical clustering of differentially expressed genes following NACT (>±2 GFOLD) finds that patients cluster by platinum sensitivity. (B) Pre and post-chemotherapy IHC (40x) demonstrating increased THBS1 (thrombospondin) expression after chemotherapy. THBS is an extracellular matrix protein whose gene expression is upregulated post chemotherapy, as highlighted in Fig.1A. Trichrome staining demonstrates stable quantity of blue staining collagen regardless of chemotherapy status. (C) Expression of core AP-1 transcription factors is consistently elevated after NACT, including c-Jun and c-Fos. (D) Treatment of OVCAR8 cells with paclitaxel, but not carboplatin, for 24 hours leads to a dose-dependent increase in c-Jun protein levels. C-Fos levels are unaffected. (E) On immunoblot Carboplatin (CP) resistant OVCAR4, TYK-nu, and PEO1 cells have elevated levels of phosphorylated c-Jun (Ser63) compared to control cells. (F) AP-1 transcriptional activity as assessed with a luciferase reporter plasmid incorporating multiple AP-1 binding sites is increased in OVCAR8 cells following treatment with paclitaxel at the indicated concentrations for 24 hours. (G) Combination of PTX with the AP-1 inhibitor SR11302 leads to synergistic cytotoxicity of OVCAR8 and TYK-nu cells at large effect sizes (ED = Effective dose; ED₇₅-ED₉₇). Combination index of less than 1 indicates synergistic activity of AP-1 inhibitor with paclitaxel. CP sens = platinum sensitive. **ANOVA p < 0.01.

Fig. 3.. Neoadjuvant chemotherapy is not associated with the acquisition of somatic mutations.

(A) Oncoprint of all somatic mutations detected in at least two patient samples. Left upper triangle in each square corresponds to pre-NACT sample and right lower triangle corresponds to post-NACT tissue. Mutational types are indicated by color. The frequency of mutations within our cohort is compared to the frequency of mutations in the TCGA analysis of HGSOC. VAF = variant allele frequency. (B) Proportion of each mutational class in all coding genes detected in the entire patient cohort, divided based on pre or post –NACT.

Fig. 4.. Recurrent copy number variations associated with neoadjuvant chemotherapy.

(A) The frequency of CNVs in the patient cohort captures recurrent CNVs observed in the TCGA analysis of HGSOC such as amplification of 3q and 8q and deletion of chromosomes 16 and 22. (B) Comparing TCGA HGSOC cohort normalized SIK2 mRNA expression to copy number alteration data reveals increased SIK2 expression with copy number gain or amplification. (C) GISTIC2.0 analysis of differentially amplified (red) and deleted (blue) regions identifies recurrent CNVs present both pre- and post-NACT, including frequent amplification of SIK2 in 11q23.1 which contains 75 total genes (Table S14). Chromosome locations are indicated on the x-axis. (D) Treatment of OVCAR8 cells with a combination of paclitaxel (PTX, 100 nM) or carboplatin (CP, 100 μM) and a SIK2 inhibitor (HG-9-91-01; 5 μM) leads to synergistic cytotoxicity based on 2-way ANOVA and Tukey’s test. Proliferation data demonstrating percent change in the Hoechst positive area between the 24 and 48 hour post treatment time points is shown in the panels on the right ***ANOVA p < 0.001. Supplementary Figures