Improving the Detection and Prevention of Ovarian Cancer

Ovarian cancer (OC) can be divided into distinct subtypes depending on the tissue of origin; surface epithelial, germ cell, sex cord stromal and metastases. Most tumours arise from the epithelial tissue of the ovary and this group is further subdivided by cell type; serous, mucinous and endometrioid, although some tumour exhibit mixed cell types (Meinhold-Heerlein et al., 2016). OC can also arise from the distal fallopian tube as shown by studies on prophylactic salpingo-oopherectomies on BRCA+ women (Kindelberger et al., 2007). The most common being the high grade serous carcinoma (HGSC), which accounts for 80% of the cases. (Oswald and Gourley, 2015) Given that clinical heterogeneity between the subtypes, this essay will focus particularly on high grade serous carcinoma of the ovary.

1. HGSC – very atypical, highly mitotic cells with papillary structure.
2. STICs are precursors to HGSC and shows similar arrangement
3. LGSC – mildly atypical with papillary structure.
4. Clear cell carcinoma – large, atypical cells showing cytoplasmic clearing.
5. Endometrioid adenocarcinoma – glandular features.
6. Mucinous – mucin rich cells and goblet cells.

Around 239,000 cases of OC were diagnosed worldwide in 2012, making it the seventh most common cancer in women. It is associated with late stage diagnosis and poor prognosis. Overall 5 year survival is 45.6% but the figure varies with stage at diagnosis. Patients at stage 1 have a survival of 92.1%, whilst patients with stage 3/4 have a 25% chance (Reid, Permuth and Sellers, 2017). This emphasises the magnitude of the survival advantage an early diagnosis gives patients and the highlights the need for sensitive diagnostic methods for early diagnosis. Incidence and risk factors. OC is more common in post-menopausal women with higher incidences in high income countries as depicted in Figure 2 (Cancer Research UK, 2012). The strongest risk factors are having the BRCA 1 (28%-44% risk) or 2 (27% risk) mutation, increasing age, family history of ovarian (5% risk) or breast cancer. Weaker risk factors are nulliparity and hormonal therapy (Justin et al, 2017).

Molecular mechanisms. In HGSC, the most common mutations are frameshift, missense and nonsense mutations of the TP53 tumour suppressor gene (Ahmed et al., 2010; Bell et al., 2011). Number gain in MDM2/MDM4 can lead to dysregulation of the p53 protein. Genomic studies have revealed that 50% of HGSCs have defects in homologous recombination, which is associated with BRCA mutations as well other alterations in other DNA repair pathways (Bell et al., 2011). The BRCA 1 and 2 proteins are involved in a scaffolding role for several other proteins such as BRIP and XRCC2, which are involved in S phase checkpoint activation and DNA repair. Mutations in BRCA1/2 lead to an accumulation of DSBs, leading to increased risk of mutations and carcinogenesis (Matulonis et al., 2016). Furthermore, molecular alterations in the PI3K, RAS-MEK, FOXM1 signalling pathways which are involved in cellular proliferation and survival have been associated with HGSC. DNA mismatch repair genes involved in the MutS pathway such as MSH2, MSH6, MLH1 and PMS2 have also been found to be mutated in OC (Matulonis et al., 2016). Symptoms and signs. OC is notorious for its non-specific symptoms (Table1) with only 1 in 400 women presenting with related symptoms get diagnosed (Goff et al., 2004). Treatment overview. High grade serous epithelial cancer of the ovaries respond well to platinum based therapy than low grade aggressive cancers. Standard care includes cytoreductive surgery followed by chemotherapy with platinum based agents and taxanes. (Matulonis et al., 2016) However, relapse rate after first cycle of treatment is high. Novel therapies such as the use of PARP inhibitors and nanoparticle based delivery system for toxic drugs such as monomethyl auristatin E are being studied (Dockery, Gunderson and Moore, 2017; Qi et al., 2017).

Cancer antigen 125 (CA125), a large glycoprotein with a single transmembrane domain, is encoded for by the MUC16 gene. Knockdown studies of the gene has showed increased bacterial adherence/invasion and decreased tight junction function, suggesting a role for mucin 16 in maintaining barrier function (Gipson et al., 2014). Furthermore, CA125 has a role in OC progression through suppression NK cells response against cancer (Patankar et al., 2005). CA125 is raised in 90% of patients diagnosed with OC (Kobayashi et al., 2008). This cost effective and accessible test is widely used to monitor treatment response using a baseline value. It is also used for diagnosis but when stage 1 cancer patients were studied, it was shown that only 50% of women had a raised CA125, making it an inefficient option for early screening of the disease. Furthermore, CA125 is also raised during physiological processes such as menstruation and in other conditions such as infections, uterine fibroids and ovarian cysts (Kobayashi et al., 2008). This non specificity of CA125 can lead to unnecessary testing. This ties in with the fact that OC is relatively rare and a high specificity is needed to have an acceptable positive predictive value for a widespread screening method.

Novel Diagnostic Techniques

The PLCO Cancer Screening trial was a large scale study that evaluated the combination of blood CA125 levels and transvaginal ultrasound as a possible screening tool for OC. The study randomised 78,216 women between the ages of 55-74 into two groups: one receiving a yearly CA125 and transvaginal USS and the other being a control. There was no statistical significance in both the rate ratio and overall mortality with similar stage distribution of cancer in both groups (Buys et al., 2005). Furthermore, it seems that combination use of CA125 and TVU produced false positive results leading to unnecessary investigation and associated severe complications. Therefore, it was concluded that this intervention is not a suitable screening method.

The UKTOCS trial evaluated the use of the ROCA algorithm to screen for OC. In the trial, 202,638 women were assigned to three groups – ROCA screening, USS screening or no screening. ROCA involved taking serial CA125 measurements, triaging the women into low, intermediate and high risk group, and following up each groups with risk weighted CA125/USS testing. The results showed no significant difference in mortality between the groups, however, median follow up was 11.1 years, and evidence showed that a longer follow up might produce a significant difference (Jacobs et al., 2016). HE4 is a small glycoprotein expressed in tissues such as the epididymis, breast and female genital tract. It has been reported to be overexpressed in early OCs as well as breast, endometrial and pulmonary malignancies. HE4 has the highest sensitivity in comparison to all other biomarkers in detecting early stage OC (Moore et al., 2008). Moreover, its specificity is 91.8% in comparison to 59.5% for CA125, an area that can be exploited to overcome the current specificity roadblock. However, this value is specific to OCs in pre-menopausal women with adnexal masses, making us question the validity of this in post-menopausal women, who make up the majority of patients (Holcomb et al., 2011). Studies on analysing urinary HE4 using mobile phone imaging and microchip ELISA at the point of care device have shown somewhat promising results but more work is needed to validate this method (Wang, Akbas and Demirci, 2015).

Lee and colleagues carried out a transcriptome analysis study using five biomarkers (AGPAT1, B2M, BASP2, IER3, and IL1B). When used in combination, it discriminated well between the papillary adenocarcinoma and healthy patients. An ROC value of 0.909 with an 85.7% sensitivity and 91.4% specificity was observed (Lee et al., 2012). Although this was a promising study, statistical power is low and further validation is needed from large scale studies for this to be used in clinical practice. In OC, there is both hypermethylation of promotor regions and hypomethylation of other CpG sites such as Alu sequences. Recent advancements in epigenetics has given potential for the use of tumour methylation pattern analysis of the circulating cell free tumour DNA in the serum to be a method for early diagnosis. Discrimination of patients with HGSC from healthy patients or those with a benign mass has been supported by statistically significant results. More importantly, when only CA125 negative samples were used, sensitivity and specificity were 90.7% and 87.5% (Widschwendter, 2017). This has potential to close the current diagnostic gap for patients with clinical symptoms and high risk, but a CA125 value less than the cut-off.

Conclusion

The heterogeneity of OC acts as a roadblock to discovering a single biomarker for screening and early diagnosis of the disease. The biggest drawback of the CA125 is that it lacks specificity. It is evident that panels of biomarkers like the transcriptome analysis and combinatory analysis of various tests like in ROCA are more specific and sensitive than single tests. The UKTOCS ROCA algorithm is more likely to be used if future studies shows significant results because the individual modalities are well established and cost effective. Although the transcriptome biomarker panel test is non-invasive, biomarkers specific to HGSC need to be identified and clinical viability is questionable because of cost and complexity. Having said this, with the rapid rate of development of molecular technology, transcriptome analysis methods may become more efficient and cheaper allowing for use within the NHS in the future. HE4 as a biomarker would only be viable in a small subset of women with ovarian cancer, making it inefficient for the overall population. Methylation arrays are already being trialled for as a commercial screening test in colorectal cancer, therefore, there is scope for the same in OC (Warton and Samimi, 2015).