Advances in screening and early detection of ovarian cancer over the past decade have included novel interpretation of serum CA125, discovery of human epididymis protein 4, which has the potential to add to CA125, and the growing understanding of the flaws of previous biomarker studies. No mortality effect was found in the ovarian screening arm of the Prostate Lung Colorectal and Ovarian Cancer Screening Trial. Concerns, however, have been raised about trial design, and the results from the UK Collaborative Trial of Ovarian Cancer Screening in the general population and other ongoing studies in the high-risk population are awaited for a definitive conclusion. Future work needs to take into account the new insights into ovarian cancer subtypes and the growing evidence that a significant proportion of ovarian cancers might originate in premalignant lesions in the distal fallopian tube.
Introduction
Ovarian cancer accounts for 4% of cancers diagnosed in women, with over 225,000 new cases diagnosed worldwide each year. Incidence rates are highest in the USA and Northern Europe and lowest in Africa and Asia. In most developed countries, it is the most common genital tract malignancy, with women having a 1–2% life-time risk of developing the disease. It is also associated with the highest mortality rates. Around 85% of cases occur over the age of 50 years, and 80–85% of cancers are epithelial in origin. The most common histological subtype of epithelial ovarian cancer (EOC) is serous ovarian cancer, which presents at advanced stages and has the poorest outcomes.
Sixty per cent of women are diagnosed at advanced stage, which has a 5-year survival as low as 10%. When the disease is caught early, 5-year survival is in excess of 90%. This forms the rationale for the premise that detecting the disease early may affect long-term survival. The ability to reduce cancer mortality through population screening is widely accepted. Cancer-specific criteria detail which cancers could most benefit from screening, and build on the World Health Organization criteria for all diseases. Cervical cancer is a good example of a gynaecological cancer fulfilling the World Health Organization criteria. Organised cervical cancer screening programmes in a number of countries across the world have reduced incidence and mortality significantly. Extensive efforts are under way to explore the possibility of influencing ovarian cancer mortality through screening.
Recent research developments with implications for ovarian cancer screening (OCS) include the following: new insights into tumour biology reflected in the proposed classification of ovarian cancers into type I; slow-growing cancers with good prognosis, such as low-grade serous, low-grade endometrioid, clear cell, mucinous and transitional (Brenner) carcinomas; and more aggressive type II, which include high-grade serous, high-grade endometrioid and undifferentiated tumours and carcinosarcomas. Future screening efforts, biomarker discovery, and refining of ultrasound indices need to ensure that both types are taken into account, and that focus on detection of type II cancers increases, as they account for most of ovarian cancer mortality. The other development with immediate effect on screening and prevention efforts results from advances in the understanding of the cell of origin of ovarian cancer. For many years, limited advances in identifying a precursor lesion to ovarian cancer have restricted the goal of screening to detect early stage disease. Growing evidence pioneered by Crum et al., however, support a model of ‘fimbrial-ovarian’ serous neoplasia, in which serous ovarian cancers start as premalignant serous tubal intraepithelial cancer lesions in the distal fallopian tube and spread to the ovary.
A wide variety of biochemical, morphological and vascular tumour markers have been explored in the context of differential diagnosis or screening with varying success. In this chapter, we focus on each of these, with particular attention to their role in screening.
Tumour markers
Tumour markers are biological substances that are produced by malignant tumours and enter the circulation in detectable amounts. They indicate the likely presence of cancer or provide information about its behaviour. In screening, the aim is to identify a marker that has high sensitivity (proportion of cancers detected by a positive test) and specificity (proportion of those without cancer identified by a negative test). Although the ideal tumour marker would have a 100% sensitivity, specificity and positive predictive value (PPV), this is not achieved in practice. The most limiting factor is lack of specificity, as most markers are tumour-associated rather than tumor-specific, and are elevated in multiple cancers, benign and physiological conditions. In most diseases, tumour markers are used for differential diagnosis and not as early detection and diagnostic tests. They have shown promise in monitoring response to treatment, detecting recurrence and predicting prognosis.
Despite massive efforts from the clinical and scientific community over the past decade, few clinically useful markers have been identified. One of the main reasons cited has been poor study design leading to inconsistent conclusions. A number of multidisciplinary groups have critically appraised the available evidence and established guidelines on best application and use of tumour markers. The National Cancer Institute Early Detection Research Network has suggested five phases for biomarker development: preclinical exploration, clinical assays and validation, retrospective longitudinal repository studies, prospective screening studies and conducting clinical randomised-controlled trials for assessing end points of cancer screening. More recently, a process of how best to evaluate biomarkers for classification or prediction has been forwarded, with the suggestion that biomarker discovery should follow the proposed PRoBE (Prospective-specimen-collection, Retrospective-Blinded-Evaluation) design.
Biochemical markers
Cancer antigen 125
Cancer antigen 125 (CA125) was first described by Bast et al. in 1981 as an antigenic determinant on a high molecular-weight glycoprotein recognised by the murine monoclonal antibody OC-125. A decade ago, it was shown to have characteristics of mucin and has been designated as MUC16. A number of CA125 assays are available, most of which correlate well with each other and are clinically reliable, but may need parallel testing if new methods are introduced. This is especially relevant in OCS using some of the newer algorithms referred to below.
CA125 is not specific to ovarian cancer and is widely distributed in adult tissues. It is found in structures derived from the coelomic epithelium (i.e. endocervix, endometrium and fallopian tube) and in tissues developed from mesothelial cells (e.g. pleura, pericardium and peritoneum). It is expressed in the normal adult ovary and in epithelial tissues of the colon, pancreas, lung, kidney, prostate, breast, stomach and gall bladder.
The level of CA125 in body fluids or ovarian cysts does not correlate well with serum levels. This is probably caused by the serum concentration being reflective of the production of the antigen by the tumour and other factors that affect its release into the circulation. The widely adopted cut-off value of serum CA125 levels of 35 kU/L is based on the distribution of values in healthy individuals, where 99% of 888 apparently healthy men and women were found to have levels below 35 kU/L. CA125 values, however, can vary widely and are influenced by age, race, menstrual cycle, pregnancy, hysterectomy and a number of benign conditions. In postmenopausal women, CA125 levels tend to be lower (<20 kU/L) than in the general population.
CA125 may also be elevated by non-gynaecological diseases causing any inflammation of the peritoneum, pleura or pericardium; pancreatitis, hepatitis, cirrhosis, ascites, tuberculosis; and other malignancies, such as pancreas, breast, colon and lung cancer. Many of the non-malignant conditions mentioned above, however, are not found in postmenopausal women, thus improving the diagnostic accuracy of elevated CA125 in this population.
Whether a raised CA125 in asymptomatic postmenopausal women is a predictor of non-gynaecological cancer is controversial. Jeyarajah et al. reported on an RCT of 22,000 women undergoing OCS, and found that elevated serum CA125 was not a predictor of a non-gynaecological malignancy on mean follow up of 2269 days. Data from the Norwegian OCS trial of 5500 women, however, showed that breast and lung cancer were over-represented among women with elevated CA125. These findings indicate that, in OCS, it is probably best to rule out other malignancies such as breast, lung and pancreas in women with rising CA125 levels with no evidence of gynaecological malignancy.
CA125 was found to be raised in 236 out of 59 (25%) stored serum samples collected from women with ovarian cancer 5 years before diagnosis. Data from a Japanese OCS study indicate that, in serous-type ovarian cancer, first elevation of CA125 to diagnosis has a shorter interval compared with those with non-serous-type (type I) disease (1.4 v 3.8 years; P = 0.011). Although 47% of non-serous-type ovarian cancers developed from slightly elevated CA125 levels between 35 and 65 kU/L, 75% serous ovarian cancers developed suddenly from a normal CA125 level (<35 kU/L). CA125 is not expressed or produced by 20% of ovarian cancers. This may partly be caused by CA125 forming circulating immune antibodies containing complexes bound to the free antigen.
The first prospective OCS study undertaken involved 4290 Swedish women aged 50 years and older. Serum CA125 using a cut-off of 30 kU/L achieved high specificity (97%) but had a low PPV of 4.6%. Since then, numerous efforts have been undertaken to improve performance characteristics of this test (see below).
Improving sensitivity, specificity and discriminatory ability of serum CA125
Specificity of screening using CA125 has been improved by adding pelvic ultrasound as a second-line test to assess ovarian volume and morphology. Jacobs et al. found that, among 22,000 women screened, multimodal screening incorporating sequential CA125 measurements and pelvic ultrasound achieved a specificity of 99.9% and PPV of 26.8% (about four operations for each cancer) for detecting ovarian and fallopian tube cancer. Since then, the interpretation of ultrasound in postmenopausal women with elevated CA125 levels has been further refined, and complex ovarian morphology is now used to define abnormality.
Novel interpretation of serum CA125 levels
Use of absolute cut-off of CA125 levels has been refined by using a more sophisticated approach and taking into account the serial values that are available in the screening context. Data from Jacobs et al. on over 50,000 serum CA125 levels involving 22,000 volunteers followed up for a median of 8.6 years showed that elevated CA125 levels in women without ovarian cancer had a flat or static profile or decreased with time, whereas levels associated with malignancy tended to rise. These data were used to construct a computerised algorithm that used an individual’s age-specific incidence of ovarian cancer and CA125 profile to estimate a woman’s risk of ovarian cancer (ROC). The closer the CA125 profile to the CA125 behaviour of known cases of ovarian cancer, the greater the ROC. The final result is presented as the individual’s estimated risk of having ovarian cancer so that a ROC of 2% implies a risk of 1 in 50. Women are triaged into low, intermediate and elevated-risk based on their ROC result. The women at intermediate risk have repeat CA125, whereas those with elevated risk are referred for a CA125 and transvaginal scanning. If either are abnormal, the women are then referred for clinical assessment with a gynaecological oncologist with a view to surgery. The full screening algorithm has been described in detail elsewhere. The ROC algorithm increases the sensitivity of CA125 compared with a single cut-off value because women with normal but rising levels are identified as being at increased risk. At the same time, specificity is improved, as women with static but elevated levels are classified as low risk. For a target specificity of 98% for preclinical detection of ovarian cancer, the ROC calculation achieved a sensitivity of 86%.
Prospective evaluation of the ROC algorithm in a randomised-controlled trial of 13,582 postmenopausal women aged over 50 years, showed a high specificity (99.8%; 95% CI 99.7 to 99.9) and PPV (19%; 95% CI 4.1 to 45.6) for primary invasive epithelial ovarian cancer. More recently, encouraging results have been reported by UK and US groups. In the prevalence screen of the ongoing randomised-controlled UK Collaborative Trial of Ovarian Cancer Screening (UKCTOCS) ( Fig. 1 ), multimodal screening using the ROC algorithm achieved a sensitivity of 89.5% (95% CI 75.2 to 97.1%), specificity of 99.8% (95% CI, 99.8 to 99.8%) and PPV of 35.1% (95% CI, 25.6 to 45.4%) with 47.1% of primary invasive epithelial cancers detected in early stage. In a smaller study of 3238 women by Lu et al., ROC algorithm followed by transvaginal scanning had a specificity of 99.7% (95% CI 99.5 to 99.9%) and PPV of 37.5% (95% CI 8.5 to 75.5%).
