Ovarian tumours, their characterisation and origins and ovarian screening

Abstract

Ultrasound assessment of ovarian tumours cannot accurately predict pathology. Recent reviews suggest that the source of many epithelial ovarian carcinomas may be the distal fallopian tube. Further study of the natural history of these tubal lesions is required before resuming ovarian screening.

Keywords::

• Gynaecological oncology

Editorial

This issue includes an article by Annaiah et al. (2012) which describes 100 post-menopausal patients who were undergoing total abdominal hysterectomy and bilateral salpingo-oophorectomy; 62 ovarian cysts or solid tumours were found (unilateral in 28 and bilateral in 17 patients, to give a total of 62 ‘tumours’). Of these 62 ‘tumours’, 34 were unilocular and all were benign pathology. Nine were multilocular and included seven benign cysts (four mucinous and three serous) and two borderline mucinous cysts, both more than 10 cm in diameter. Among 8 uni- or multilocular tumours with solid components were two borderline tumours (one mucinous and one serous) and one mature cystic teratoma containing a squamous cell carcinoma. There were 11 solid tumours, of which two were benign Brenner tumours, five fibromas and four malignant tumours; two granulosa cell tumours, one metastatic colorectal and one metastatic endometrial adenocarcinomas. Of the 30 tumours < 5 mm in diameter, only one was malignant and that was a granulosa cell tumour. These findings raise questions about whether ultrasound can predict pathology and how to manage patients who are found coincidentally to have ovarian tumours, e.g. when undergoing scanning to assess endometrial thickness when bleeding occurs while using hormone replacement therapy.

There are a growing number of articles that describe findings from ovarian screening programmes. One of the longer running series has been in Kentucky, USA, which started in 1987, used annual transvaginal ultrasound scanning and continued until 2005 (DePriest and DeSimone 2003; van Nagell et al. 2007). A total of 25,327 women underwent screening and 364 (1.4%) had persisting ovarian abnormality; these women underwent CA125 estimation, tumour morphology indexing (ovarian volume, cyst wall and septation) and Doppler flow sonography, prior to laparoscopy or laparotomy. Among those surgical specimens were 44 (14%) with 35 adenocarcinomas and nine serous tumours of low malignant potential. Additionally, seven metastatic tumours from the appendix, peritoneum, breast and endometrium were found. However, a significant number of women underwent surgery for what was likely to be non-significant pathology. In an update of that study, McDonald et al. (2010) reviewed 395 patients undergoing surgery from 2001–2008; 118 had ovarian cancer, 13 had borderline ovarian tumours and 264 had benign ovarian tumours. The positive predictive value was 84.7% if the adnexal mass was complex or had solid morphology and the serum CA125 was > 35 U/ml. These findings identified 98.6% of patients with stage III and IV disease but only 77.3% of those with stage I or II ovarian cancer.

Kobayashi et al. (2008a) reported on a screening programme conducted in Japan between 1985 and 1999. Asymptomatic post-menopausal women were screened using annual pelvic ultrasound examination plus CA125. Among the 41,688 screened women were 27 detected cancers, but eight more were diagnosed outside the screening programme. There were 32 women among the 40,779 control or unscreened group who developed ovarian cancers. The stage of cancer at diagnosis or detection was more likely to be stage I in the screened women, but this was not significantly different to the controls. In a further analysis (Kobayashi et al. 2008b), 4,859 had an abnormal transvaginal ultrasound scan but a CA125 < 35, of which 4,741 had simple morphology and 118 had more complex appearances. Surgery was performed on 912 with simple, and 69 with complex appearances and ovarian cancer was identified in eight women, three of whom were more than stage I. Again the point must be made that many women underwent surgery (more than 100 procedures:each cancer detected) and that cancers will be found among those women with abnormal scan appearances but non-elevated CA125.

Partridge et al. (2009) described the experience of the PLCO (prostate, lung, colorectal and ovary) Cancer Screening Trial, which included 34,261 women, 1,170 of whom underwent surgery to find 60 (5.1%) ovarian cancers, i.e. a ratio of cases operated on to cancers detected of 19.5:1. Not only was there a very high ratio of cases operated on to cancers detected, but 72% of those cancers detected by screening were stage III or IV, and thus, survival was not improved by earlier detection. Additionally, there were 29 cases of ovarian cancer that were missed by screening but subsequently discovered.

Menon et al. (2009) have published some interim results of the initial screening in the UK Collaborative Trial of Ovarian Cancer Screening (UKCTOCS). Between 2001 and 2005, 202,638 postmenopausal women were recruited into the project. One quarter (50,640) underwent annual CA125 screening with transvaginal ultrasound as a second-line test (multimodal screening); 97 subsequently underwent surgery to identify 42 primary ovarian or tubal cancers, but five women developed a relevant cancer during the year after the screen. Among the 48,230 women who had ultrasound screening alone, 845 underwent surgery to identify 45 relevant cancers, but eight more were missed and were detected during the year after screening. Among the cancers identified were 28 borderline ovarian tumours; eight in the multimodal and 20 in the ultrasound screened groups.

A recent Green-top Guideline from the Royal College of Obstetricians and Gynaecologists (RCOG 2011) on the management of ovarian masses in premenopausal women stresses that CA125 is unreliable in differentiating benign from malignant ovarian masses in such women, and is raised in only 50% of early stage disease. Other tumour markers including AFP, hCG and LDH (lactate dehydrogenase) should be measured in premenopausal women with complex appearing masses because of the increased likelihood of non-epithelial tumours. The search for better markers, e.g. HE4, for epithelial ovarian cancers continues (Moore et al. 2010).

Our thinking about the origins of ovarian cancer has been confused because we have often regarded epithelial ovarian carcinomas (EOCs) as a single pathology, and adhered to Fathalla's theory that EOCs are caused by the recurring trauma to the ovarian surface epithelium from ovulation (Fathalla 1971). Booth et al. (1989) in a hospital-based case–control study of 235 women with ovarian cancer in London and Oxford hospitals, reported that high gravidity, hysterectomy, female sterilisation and oral contraceptive use reduced the risk of ovarian cancer, while infertility and late age at menopause were associated with an increased risk; one deduced that pregnancy and suppression of ovulation reduced damage to the ovarian surface, while infertility led to treatments including ovulation induction, which increased the risk. This was further supported by publications from Whittemore et al. (1992) and Rossing et al. (1994) in which the use of gonadotrophins and clomiphene were implicated. Subsequent analyses (Venn et al. 1995, 1999; Klip et al. 2000) including one with the editor's collaborators (dos Santos Silva et al. 2009) could not find an increased incidence or mortality from ovarian cancer among women with infertility or after ovarian/ovulation stimulation, although a most recent report from the Dutch group (van Leeuwen et al. 2011) describes an increased risk of borderline and invasive ovarian tumours after ovarian stimulation for in vitro fertilisation.

Much effort has focussed on the surface epithelium of the ovary (a fragile monolayer of cells), and the inclusion cysts invaginated within the ovary, to detect pre-invasive neoplastic or dysplastic changes (Plaxe et al. 1990; Mittal et al. 1993; Resta et al. 1993; Deligdisch et al. 1999). Nieto et al. (2001) identified higher dysplasia scores in women who had undergone ovulation induction compared with fertile controls, and linked such dysplastic changes with increased expression of p53 (Nnene et al. 2004). Continuing associations were made between ovarian cancer and infertility (Nieto et al. 1999; Hardiman et al. 2000), although attempts to link the cancer to infection, and organisms known to be linked with cancer or infertility, were unrewarding (Hardiman et al. 2000; Wong et al. 2007).

However, there is now growing evidence that at least some ovarian cystadenocarcinomas (and particularly the most commonly found high-grade serous type) arise from distal tubal epithelium (Crum et al. 2007; Rabban et al. 2010; Seidman et al. 2010). Molecular differences between high- and low-grade serous tumours include p53 mutations (Vang et al. 2009), and the detection of p53 mutations as well as intraepithelial carcinomas in the distal tube of women undergoing prophylactic salpingo-oophorectomy because they carry breast cancer (BRCA) mutations, is described in Crum and colleagues’ review article (2007). Their article concludes that these early pathological changes in the distal tube are unlikely to be detected by biomarkers, and from their microscopic beginnings are not going to be recognised early by ultrasound screening.

The findings of Annaiah et al. (2012) of borderline ovarian tumours raise the question whether invasive epithelial ovarian carcinomas arise from borderline tumours and if so, what is their natural history. They were reported by Rossing et al. (1994) among women with ovarian tumours following treatment with clomiphene, and the review by Ness et al. (2002) of eight case–control studies demonstrated an increased incidence of borderline tumours among nulligravid women who had used fertility drugs. The study by van Leeuwen et al. (2011) reported finding borderline tumours in women who had undergone IVF. The report by Menon et al. (2009) suggests these tumours are more likely to be detected among women undergoing ultrasound screening as compared with those in their multimodal screening group. Are these tumours significant; are they important to detect and remove? It would appear that the implications of ovarian screening have opened up new avenues to explore the origins and characteristics of ovarian cancers, and until we understand more about these areas, we should defer screening women for such cancers and subjecting them to unnecessary surgery.

Post-script

Since preparing this Editorial, BJOG has published an article by Ahmed et al. (2012) that should be brought to the attention of readers. (Ahmed AA, Becker CM, Bast RC. 2012. The origin of ovarian cancer. BJOG 119: 134–136.)

Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.