Liquid Biopsies for Improving the Pre-operative Diagnosis of Ovarian Cancer

OVI-DETECT Liquid Biopsies for Improving the Pre-operative Diagnosis of Ovarian Cancer

An accurate preoperative diagnosis of an ovarian tumor is important for the patients' surgical work-up, proper referral to oncological centers and for the patients' mental wellbeing since uncertainty about the nature (benign vs malignant) of an ovarian tumor may cause anxiety. Currently, the Risk of Malignancy Index (RMI), with a cut-off value of 200, is often used in the Netherlands to select patients with an increased risk of ovarian cancer that should be referred to an oncologic center. However sensitivity and specificity of the RMI-score are far from optimal. Around 40% of the referred patients have benign disease in final pathological examination. Therefore, other models have been developed, such as the IOTA (International Ovarian Tumor Analysis) consortium algorithms, but these models require training, expertise and are subjective. To determine the nature of an ovarian tumor, histological examination is the golden standard. However, a pre-operative biopsy of an ovarian tumor is undesirable because of the risk of spill of tumor cells in the abdominal cavity. Therefore, there is an urgent need for non-invasive diagnostic tools to determine the nature of an ovarian tumor pre-operatively. Liquid biopsies could be such a non-invasive tool. Currently, circulating tumor DNA (ctDNA) circulating tumor cells (CTC), microRNA (miRNA) and tumor-educated platelets (TEPs) are available and can function as a potential blood-based biosource for (early) cancer diagnostics. Previous studies show promising results of liquid biopsies are used in (early) detection of cancer, also for ovarian cancer.

Therefore, a diagnostic algorithm will be developed using ct-DNA and TEPs as liquid biomarkers in combination with the existing ultrasound models (RMI and IOTA-models) and tumor markers (CA125 and HE4) to differentiate between early ovarian cancer and benign ovarian tumors pre-operatively.

Nature and extent of the burden and risks associated with participation, benefit and group relatedness. There is no extra burden/risk for the patients in this study. Five extra vials of blood will be collected from each participant and two questionnaires will be filled out.

In the Netherlands 7600 women are annually diagnosed with an ovarian tumor. Only 5% of these tumors are malignant in the final histology assessment. This means that a general gynecologist is confronted with a patient with low stage ovarian cancer less than once a year. Therefore an accurate preoperative diagnosis of an ovarian tumor is challenging. This accurate pre-operative diagnosis is important because patients with an ovarian carcinoma have to undergo extensive surgery in an oncology center, performed by a gynecologist-oncologist.

The current preoperative possibilities to distinguish benign and malignant ovarian tumors are based on classification systems containing clinical, biochemical and ultrasound characteristics. For example, predictive ultrasound models developed by the IOTA (International Ovarian Tumor Analysis) consortium have been widely used. However, these models require training and expertise and are therefore not always easy to implement. The predictive value of current serum-biomarkers such as CA-125 is limited because this biomarker is not increased in 50% of early stage ovarian carcinoma and CA-125 may also have increased in benign gynecological conditions such as endometriosis.

Current Dutch guidelines make use of the Risk of Malignancy Index (RMI) to determine whether the risk of ovarian carcinoma is increased. This score is based on the concentration of CA125, specific ultrasound characteristics and menopausal status.

According to Dutch guidelines, patients with ovarian tumor are referred to oncology centers if the RMI is increased (>200). The published sensitivity and specificity of RMI in a non-selected population of patients with ovarian tumors is 72% and 92%, respectively. However, the population treated in the oncology centers is enriched with patients with RMI >200. In this selected population, our own prelimenary data show that the sensitivity is 84% for RMI and the specificity is only 51%. This means that the incidence of malignancy within this population is 40%. This is unacceptably low because this implies that half of the patients with benign tumors referred to oncology centers undergo unnecessarily extensive surgery and these patients become unnecessarily emotionally burdened with the possibility of getting cancer.

Tissue biopsies are an important tool in the treatment of ovarian carcinoma because theses procedures can confirm or rule out the presence of a malignancy preoperative. At an early stage, however, tissue biopsy is considered as an unwanted invasive procedure, as this procedure can cause tumor spreading and has an invasive character for patients.

In short, despite the development of numerous prediction models, accurate preoperative diagnosis of early stage ovarian carcinoma is a challenge. There is an urgent need to develop prediction models with a high degree of accuracy that are easy to implement in the clinic to maximize the number of malignant tumors treated in oncology centers.

Blood-based biopsies, 'liquid' biopsies, as an alternative to traditional tissue biopsies, are emerging as these biopsies can provide accurate and comprehensive information on tumors. Examples of such include circulating tumor DNA (ctDNA) circulating tumor cells (CTC) and tumor-educated platelets (TEPs). These platelets are normally responsible for hemostasis, but appear to be able to include tumor signals found in the presence of mirco-tumor RNA. By using whole genome sequencing and detecting structural DNA and RNA changes in the ctDNA and TEPs, malignancies can be detected or excluded. These DNA changes were firstly demonstrated by coincidence when using the NIPT (non-invasive pregnancy test), which, in addition to assessing possible errors in fetal DNA, is also able to detect abnormalities in maternal DNA; in asymptomatic pregnant women, DNA changes were found in maternal DNA accounting for the presence of a malignancy.

Research shows that within patients with early stage ovarian carcinoma (st I and II), ctDNA analysis showed a sensitivity of 69% and a specificity of >99%. The amount of ctDNA correlates with the tumor burden. Other studies conducted within small patient populations show that when ctDNA is combined with existing tumor markers, CA-125 for example, sensitivity and specificity both increase. From prelimanery data on TEP-analysis performed in the NKI-AvL a sensitivity of 76% and specificity of 98% for high-grade ovarian carcinoma was found and a sensitivity of 81% and specificity of 80% for low-grade carcinoma was found.

Therefore, this study investigates the clinical value of longitudinal assessment of liquid biopsy-derived information for preoperative diagnosis in patients suspected of early ovarian cancer

Inclusion Criteria:

1. Age ≥18 years

2. Presence of a ovarian tumor and referred to specialized center for surgery based on:

   1. Any ultrasound model e.g. RMI-scoring model ; IOTA-rules
   2. Subjective assessment of the referring gynecologist
   3. Normal Glomerular Filtration Rate (GFR): >60ml/min/1,73m2

3. General criteria: a. Understanding of Dutch language b. Fit for surgery (WHO 1-2) c. Written informed consent

Exclusion Criteria:

1. Suspicion of advanced-stage of disease, e.g. ascites or peritoneal depositions
2. History of cancer (excl. BCC) within 5 years prior to inclusion
3. Multiple malignancies at the same time