Current developments in ovarian cancer screening

Abstract

Over 150 delegates from the UK, USA and Europe with a core interest in risk prediction and screening for ovarian cancer attended the International Conference on Ovarian Cancer Screening held on 29–30 November 2011 in London, UK. The scientific program was driven by two experts in the field – Usha Menon and Ian Jacobs – with assistance from the scientific committee, which included Steve Skates, Jatinderpal Kalsi, Anna Lokshin, Uzi Beller, Tim Mould and Ranjit Manchanda. Over the 2 days, key opinion leaders and researchers reported on the latest developments, and debated the future of risk prediction and screening for ovarian cancer.

Keywords::

• biomarkers
• genes
• imaging
• ovarian cancer
• risk prediction
• risk stratification
• screening

The conference started with a welcome from Ian Jacobs (University of Manchester, Manchester, UK), followed by an overview of the current management of ovarian cancer (OC) by Tim Mould (University College Hospitals NHS Trust, London, UK). Stuart Campbell (Create Health Ltd, London, UK), a stalwart in the field of pelvic ultrasonography, chaired the first session on differential diagnosis in symptomatic patients. Robert Bast (University of Texas MD Anderson Cancer Center, TX, USA), who in the early 1980s discovered CA125, discussed recent biomarker panels and algorithms (Risk of Malignancy Index [RMI], Risk of Ovarian Malignancy Algorithm [ROMA] and OVA1) for OC diagnosis. His view was that the current challenge, especially in the USA, was implementation so that OC patients could be operated on by trained gynecological oncologists. Of the many OC markers, CA125 and HE4 provided the greatest discrimination between malignant and benign adnexal masses, with the latter particularly useful in premenopausal women. Similar results were reported by Estrid Hogdall (Herlev Hospital, Univerity of Copenhagen, Denmark) in the Danish Pelvic Mass Study of 1282 consecutively enrolled patients. Equally, transvaginal ultrasound (TVS) can separate malignant and benign pelvic masses, through the incorporation of a morphology index (Edward Pavlik, University of Kentucky Medical Center, KY, USA), or the use of simple rules developed by the International Ovarian Tumor Analysis group (Dirk Timmerman, University Hospitals Gasthuisberg Campus, Belgium).

The next session focused on identification of women at risk of developing OC. At present, risk stratification is based on age, family history of breast/ovarian cancer, or mutations in the BRCA1/2 genes. Simon Gayther (University of Southern California, CA, USA) provided a stimulating overview of the genetics of OC. Based on current literature, approximately 22% of OC risk is attributable to heritable factors. High-penetrance genes eliciting current interest include BRIP1, RAD51C and RAD51D. Although the risk associated with them may not be as strong as with BRCA1/2, it is nonetheless substantial. In breast–ovarian cancer families, mutations in RAD51C and RAD51D conferred a sixfold increased risk of OC, but little or none for breast cancer (Clare Turnbull, Institute of Cancer Research, London, UK). In the general population, Gayther described the genome-wide analysis studies that have identified six new susceptibility loci [1]. Paul Pharoah (University of Cambridge, Cambridge, UK) estimated that 5% of women at highest risk in the general population may have a lifetime risk of OC of 3.7%, based on preliminary data from three population-based OC case–control studies. As many more genetic loci are identified, a combination of genes and lifestyle risk factors may identify a subset of the general population at sufficiently high risk to warrant risk reducing salpingo-oophorectomy.

Ephrat Levy-Lehad (Shaare Zedek Medical Center, Jerusalem, Israel) reported on an Israeli study, where BRCA1/2 families were identified by testing healthy Ashkenazi Jewish men. She concluded that many carriers, who would be missed by current family history-based testing, would be identified through population screening. Similar findings were reported by Ranjit Manchanda (University College London, London, UK) in the UK Ashkenazi Jewish population. In addition, in the pilot randomized controlled trial Genetic Cancer Prediction Through Population Screening (GCaPPS) comparing family history versus population-based approaches to BRCA1/2 testing, there were no differences in short-term psychological/quality of life outcomes.

Epidemiological factors influence OC risk. Susanne Krüger Kjær (Danish Cancer Society, Copenhagen, Denmark) provided an excellent overview of lifestyle and hormonal/reproductive risk factors, and reported that Type I and II OC are likely to be associated with different risk factors. A Danish case–control study (Mette Tuxen Faber, Danish Cancer Society, Copenhagen, Denmark) showed that intake of dairy products was associated with a modest increase in OC risk.

The afternoon session on 29 November was devoted to updates and insights from recent/ongoing ovarian cancer screening (OCS) trials. Christine Berg (National Cancer Institute, WA, USA) presented the results of ovarian screening in the Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening trial. Simultaneous annual screening with CA125 and TVS compared with usual care in 78,216 US women did not reduce OC mortality, and was associated with complications in those who underwent surgery for false-positive results [2]. These findings have given the gynecological oncology community pause for thought, and have led some researchers to doubt the utility of current OCS strategies. Usha Menon described the design of the United Kingdom Collaborative Trial of Ovarian Cancer Screening (UKCTOCS) trial, the largest to date involving 202,638 postmenopausal women [3], and highlighted how this differed from the PLCO trial [4]. Two screening strategies, one using serum CA125 as the primary test, and the other using TVS, were compared to no screening. Central to the CA125 screening strategy was the use of serial trends over time, assessed by the Risk of Ovarian Cancer Algorithm (ROCA) [5], described in detail later by Steve Skates (Harvard Medical School, MA, USA). Menon also presented unpublished data on a small cohort of women from UKCTOCS, who were detected by the ROCA to be at increased risk despite normal scans. These data suggested that small-volume disease was detectable by measuring sequential CA125 change using ROCA. Bast also reported on the successful use of ROCA in a small US single-arm screening trial in low-risk postmenopausal women. Jacobs explored lead time of screening, which he estimated to be 1.5–1.8 years for CA125 screening using ROCA.

Pavlik spoke on attitudes towards OC screening in the Kentucky Ultrasound Screening Study [6]. Although women realized that OC was less common than breast cancer, and that the effectiveness of screening was unproven, they still wished to undergo screening, especially if a relative had died of the disease. Elisabetta De Matteis (Università degli studi di Modena e Reggio Emilia, Italy) stated that screening may provide a means of coping with cancer risk-related worry.

Skates reported on the use of ROCA in the US high-risk OCS trials under the auspices of the Cancer Genetics Network and Gynaecological Oncology Group. Adam Rosenthal (Queen Mary, University of London Medical School, London, UK) reported on the largest high-risk OCS study, UK Familial Ovarian Cancer Screening Study (UKFOCSS). In this single-arm prospective study, women were initially screened using annual serum CA125 (using a cutoff) and TVS. Since 2007, Phase II trials have involved 4-monthly CA125 screening, interpreted using the ROCA and annual TVS.

The second day of the conference started with a focus on the natural history of OC, chaired by Martin Widschwendter (University College London, London, UK), who spoke on the role of epigenetics in OC development. There is much debate regarding the cell of origin of OC, and Christopher Crum (Harvard Medical School; Brigham and Women’s Hospital, MA, USA) provided an excellent overview of the Fallopian tube as an important site for the initiation of pelvic serous cancer. The fimbrial mucosa harbors some of the earliest possible lesions: serous tubal intraepithelial carcinoma and secretory cell outgrowths [7]. In a case–control study, serous tubal intraepithelial carcinomas were reported to have the shortest telomeres, followed by tubo–ovarian dysplastic lesions from women with BRCA1 mutations (Gautier Chene, Centre Jean Perrin, France).

Using published data from occult serous cancers detected by risk-reducing salpingo-oophorectomy in BRCA1/2 mutation carriers and mathematical modeling [8], Patrick Brown (Stanford University School of Medicine, CA, USA) postulated that the median diameter of a serous OC is approximately 3 cm when it progresses to stage III, and that it is not possible with existing technologies to detect OC early enough to impact on mortality. Truly cancer-specific biomarkers, or development of novel approaches beyond traditional blood protein biomarkers, are required. Brown described ESRRA–C11orf20, a recurrent gene fusion that may be suitable for detecting early serous OC.

The last session of the meeting on OC biomarker discovery was chaired by Sudhir Srivastav (National Cancer Institute, MA, USA). Skates described the development of ROCA [5] from the mid-1990s [9], initially using serial CA125 data from the Stockholm [10] and then the early Barts [11,12] OCS trials. The algorithm calculates a woman’s OC risk by comparing the woman’s CA125 profile with that of cases and controls. Anna Lokshin (University of Pittsburgh, PA, USA) highlighted the need to identify novel biomarkers that can detect preclinical disease, and suggested that testing for urinary biomarkers may represent a more robust alternative. Nitzan Rosenfeld (University of Cambridge, Cambridge, UK) described an alternative assay that currently detects patient-specific TP53 gene mutations in circulating tumor DNA. Further development that allowed direct sequencing of TP53 circulating tumor DNA could hold promise for early detection. Joachim Schultze (University of Bonn, Bonn, Germany) reported on a case–control study of blood transcriptomics in OC. Using mathematical simulation and adaptive learning to predict outcome in a larger patient cohort, he concluded that this approach was likely to be useful in OCS. Charles Drescher (Fred Hutchinson Cancer Research Center, WA, USA) reported that HE4 had higher sensitivity than TVS for detecting all stages of high-grade serous tumors, but lower sensitivity for early-stage/low-grade tumors, in a case–control study using samples from the PLCO trial.

The afternoon of the second day was devoted to a workshop titled ‘Quo Vadis? – Biomarker discovery for Ovarian Cancer Screening’, chaired by Uzi Beller (Shaare Zedek Medical Center, Jerusalem, Israel). Menon summarized the key questions and challenges, and called for collaborative efforts, as the use of appropriate sample sets is critical for biomarker discovery. Jacobs highlighted the need to use preclinical samples that predate OC diagnosis for discovery. An industry perspective on developing a diagnostic test was provided by Nicholas Hoyle (Roche Diagnostics GmbH). Srivastav described the ProBE study design, developed by the Early Detection Research Network investigators for identifying disease-specific biomarkers [13]. Srivastav urged the community to set up a network for sharing samples and data, in order to facilitate OC biomarker discovery. There was agreement by most participants that such an alliance, along the lines of the highly successful Ovarian Cancer Association Consortium, would be worthwhile.

Financial & competing interests disclosure

U Menon has a financial interest in UCL Business and Abcodia Ltd in the third party exploitation of clinical trials biobanks, which have been developed through the research at UCL. U Menon has received grant funding from the MRC, CR-UK, NIHR and Eve Appeal for research into ovarian cancer screening. The conference was organised by Kenes Group. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

No writing assistance was utilized in the production of this manuscript.