Functional genomics of ovarian cancer
Our laboratory focuses on discovering improved treatments for epithelial ovarian cancer using laboratory and clinical studies.
Ovarian cancer has a high healthcare burden because of low cure rates and frequent recurrent disease that causes significant symptoms for patients. This is despite the fact that ovarian cancer is initially sensitive to systemic treatments and most patients are free of disease after completing initial surgery and chemotherapy. The fundamental problem that we are addressing is to understand how ovarian cancer cells escape initial treatment and the molecular mechanisms by which they acquire resistance to further therapy. Using genomic and functional studies we are identifying new biomarkers and treatment targets for testing in clinical trials.
Genomic studies of chemotherapy response in vivo
To identify genetic alterations that are selected for during the acquisition of drug resistance we are carrying out prospective clinical studies that collect cancer samples before and during neoadjuvant treatment. Our initial studies have focused on the drugs carboplatin and paclitaxel as these are the most important therapies in ovarian cancer. By using expression analysis and bioinformatics methods that have been developed to model the acquisition of resistance, we have identified clinically relevant biomarkers that overlap with independently identified genes from RNA interference screens (Swanton et al., Cancer Cell 2007; 11: 498).
Our studies depend upon having homogeneous patient cohorts with similar clinical characteristics. However, response to treatment in tumour masses can be heterogeneous and mixed response frequently occurs at different anatomical sites. For example, primary ovarian masses may respond better than peritoneal metastases. This differential response may be a result of variable blood supply and hypoxia that limits delivery and efficacy of chemotherapy.
We have confirmed these observations using functional magnetic resonance imaging for perfusion (Sala et al., Eur Radiol 2010; 20: 491) and diffusion and are now using imaging data to target the collection of tissues from responding and non-responding areas. This will allow us to calibrate genomic profiles more precisely and to better identify the molecular determinants of resistance. High throughput sequencing with Illumina technologies is being used to quantitate expression and genomic changes and to identify novel fusion transcripts and mutations (Figure 1).
Figure 1. High-resolution array CGH analysis of ovarian tumours. The Illumina 1M SNP array gives (A) Copy number data and (B) SNP allele calls. Applying the QuantiSNP segmentation algorithm provides (C) copy number calls. Blue, four copies; Green, three copies; Yellow, one copy. We are using this along with high-throughput sequencing data to characterise ovarian tumour heterogeneity and evolution towards chemotherapy resistant disease.
Differential sensitivity to paclitaxel as compared to carboplatin may depend on cellular pathways involved in maintaining chromosomal stability (CIN). To ask whether this may be clinically relevant we have tested surrogate expression markers of CIN in samples from a prospective neoadjuvant study and have shown that high measures of CIN predict resistance to paclitaxel and increased sensitivity to carboplatin (Swanton et al., PNAS 2009; 106: 8671) (Figure 2). Thus, measuring CIN pre-treatment may optimise choice of treatment for patients.
Figure 2. Expression of CIN70 genes determines sensitivity to paclitaxel and carboplatin. The figure contrasts basal median gene expression for each CIN70 gene in tumours with differing responses to paclitaxel and carboplatin. Paclitaxel-resistant tumours exhibited a higher median log‑intensity of the CIN70 signature compared with paclitaxel-sensitive tumours (P=0.043). CIN70 gene expression differed signiﬁcantly between tumours subsequently resistant to paclitaxel and tumours resistant to carboplatin (P=0.044; Student 2-sided t-test).
The key oncogenic and tumour suppressor genes for high-grade ovarian serous carcinoma have not been identified as this type of tumour has high rates of genomic instability, where many of the described alterations may be passenger mutations. Numerous studies have tested the association between TP53 mutations in ovarian cancer and prognosis but these have been consistently confounded by limitations in study design, methodology and/or heterogeneity in the sample cohort. To identify the true prevalence of TP53 mutations in high-grade pelvic serous carcinoma, we sequenced exons 2–11 and intron-exon boundaries in tumour DNA from 145 patients with invasive serous carcinoma of the ovary, fallopian tube and primary peritoneal cancer. Surprisingly, pathogenic TP53 mutations were identified in 97% (n = 119/123) of HGS cases (Ahmed et al., J Pathol 2010; 221: 49). This is the first comprehensive mapping of TP53 mutation rate in a homogeneous group of high-grade pelvic carcinoma patients and shows that mutant TP53 is a driver mutation in the pathogenesis of HGS cancers.
Mechanisms of taxane resistance and the role of extracellular matrix
Taxanes, such as paclitaxel, interfere with the dynamic growth of microtubules by directly binding to them, leading to mitotic arrest and apoptosis. Paclitaxel is widely used to treat ovarian and breast cancers but drug resistance limits its clinical usefulness to only half of patients who receive it.
Alterations in the ratio of tubulin isoforms or mutations in tubulin can alter microtubule stability and sensitivity to taxane drugs. By studying cell line models of taxane resistance along with clinical samples we have recently shown that loss of the ECM protein, transforming growth factor beta induced (TGFBI), was sufficient to induce paclitaxel resistance in cells and ovarian cancer tissues (Ahmed et al., Cancer Cell 2007; 12: 514). We have also shown that TGFBI induces microtubule stabilisation that is dependent upon integrin-mediated FAK and RHO signalling pathways. Extracellular matrix proteins have been implicated in the acquisition of drug resistance in ovarian cancer although the mechanism by which this is achieved is unclear. Loss of TGFBI induces resistance by altering microtubules which are the direct pharmacodynamic target of paclitaxel. This work shows that the effects of ECM proteins on drug resistance may be very specific to particular cytotoxic treatments. As 30% of ovarian cancers do not express TGFBI, it may be an important biomarker for paclitaxel response.
Current projects are characterising how TGFBI interacts with integrins and other cell surface receptors and how this may be modulated therapeutically. It is now clear that TGFBI exerts its effects specifically through beta-3 integrins but is also co-regulated, and interacts with, other ECM proteins implicated in drug resistance. To identify the downstream pathways from FAK and RHO that alter microtubule stability, we have generated knock-out somatic cell lines using homologous recombination. These knock-out models have provided a powerful system to identify microtubule associated proteins responsible for effects on paclitaxel resistance. As TGFBI has complex roles in organising interactions between cells and ECM, we have studied its function in early development in Xenopus to identify how it may affect cell migration. Both loss and gain of function experiments have shown that TGFBI is required for somite development in Xenopus.