Background: Ovarian cancer survival is stage-dependent: Stage I patients have 90% 5-year survival versus 15% for stage IV. Over 70% of patients worldwide are diagnosed at advanced stages. Ovarian cancer presents with non-specific symptoms (abdominal bloating, early satiety, discomfort/pain, bowel/urinary changes). Current National Institute for Health and Care Excellence guidelines recommend that symptomatic women presenting to primary care are tested with cancer antigen 125 and ultrasound, then referred to secondary care for further triage if these tests are abnormal. Current standard of care risk prediction model used to triage women in National Health Service secondary care is Risk of Malignancy Index 1 combining cancer antigen 125 and simple ultrasound features, which at 250 threshold has 70% sensitivity and 90% specificity. Newer models offer potential for improved sensitivity, earlier diagnosis and better survival outcomes.
Objectives: To evaluate diagnostic strategies for ovarian cancer in women with non-specific symptoms through systematic review, United Kingdom Collaborative Trial of Ovarian Cancer Screening data set analysis, prospective studies and health economic evaluation comparing Risk of Malignancy Index 1 against newer approaches including Risk of Ovarian Malignancy Algorithm, Ovarian-Adnexal Reporting and Data System and International Ovarian Tumour Analysis models, including International Ovarian Tumour Analysis Assessment of Different NEoplasias in the adneXa.
Methods: Four concurrent work packages: (1) Cochrane systematic review; (2) United Kingdom Collaborative Trial of Ovarian Cancer Screening data set model development; (3) prospective multicentre diagnostic accuracy study (ROCkeTS) with parallel pre/postmenopausal cohorts; and (4) cost-consequence analysis. Allied analyses investigated psychological impact and cancer outcomes from symptom-triggered pathways. ROCkeTS recruited 2453 women across 23 hospitals (2015-23) with symptoms, raised cancer antigen 125 and/or abnormal imaging. Women completed questionnaires, donated blood and underwent transvaginal ultrasound scored by International Ovarian Tumour Analysis terminology by certified National Health Service sonographers with quality assurance. Reference standard was histology for surgical cases or 12-month wellbeing ascertainment. Primary outcome: primary invasive ovarian cancer versus benign or normal.
Results: The Cochrane systematic review (58 studies, 30,121 patients and 9061 ovarian cancer cases) demonstrated that most published diagnostic test accuracy studies failed to differentiate between pre- and postmenopausal women, and all were conducted in high-prevalence settings, limiting applicability to routine practice. In the ROCkeTS prospective study in premenopausal women, in the initial cohort recruited prior to protocol change (n = 857), Risk of Malignancy Index 1 at threshold 250 showed poor sensitivity (42.6%, 95% confidence interval 28.3 to 57.8) but high specificity (96.5%, 95% confidence interval 94.7 to 97.8). All other tests improved sensitivity but dropped specificity. International Ovarian Tumour Analysis Assessment of Different NEoplasias in the adneXa at 10% threshold achieved significantly higher sensitivity (89.1%, 95% confidence interval 76.4 to 96.4), higher than all other tests with acceptable specificity (73.2%, 95% confidence interval 69.9 to 76.4). In the ROCkeTS prospective cohort study in postmenopausal women (n = 1242), Risk of Malignancy Index 1 at 250 demonstrated better performance (82.9%, 95% confidence interval 76.7 to 88.0), but International Ovarian Tumour Analysis Assessment of Different NEoplasias in the adneXa at 10% had the best sensitivity at 96.1% (95% confidence interval 92.2 to 98.4) compared to Risk of Malignancy Index 1 with the least drop of specificity. Risk of Ovarian Malignancy Algorithm at manufacturer recommended threshold and Ovarian-Adnexal Reporting and Data System did not improve on Risk of Malignancy Index 1 sensitivity in postmenopausal women. Cancer prevalence differed between premenopausal (5.7%) and postmenopausal (17%) cohorts. Early-stage cancer (I/II) were diagnosed in 60.2% of premenopausal and 41% of postmenopausal cohorts. Cancer diagnosis rates were very low (1.6%) in women under 40 years. High anxiety and distress were noted, particularly in younger women. One in four women with high-grade serous ovarian cancers were diagnosed at early stage (I/II). Complete cytoreduction was achieved in 61.3% of cases, with optimal cytoreduction (≤ 1 cm residual disease) in an additional 15.1%. Cost-consequence analysis demonstrated that a two-step strategy deployed at the same ultrasound sitting, initially triaging out benign looking tumours on ultrasound, then calculating ovarian cancer risk with International Ovarian Tumour Analysis Assessment of Different NEoplasias in the adneXa ultrasound model at 10% demonstrated the best balance across cost, diagnostic yield and cancer deaths compared to other diagnostic strategies.
Limitations: Cohort study required key changes to protocol and post-pandemic recruitment was slow.
Conclusions: International Ovarian Tumour Analysis Assessment of Different NEoplasias in the adneXa ultrasound at 10% threshold, delivered by trained National Health Service sonographers demonstrated superior diagnostic performance compared to Risk of Malignancy Index 1 and should be considered as new standard of care for suspected ovarian cancer in pre- and postmenopausal women. A two-step strategy using International Ovarian Tumour Analysis Assessment of Different NEoplasias in the adneXa offers optimal balance across cost, diagnostic yield and cancer death reduction. Implementation requires sonographer training investment and quality assurance.
Future research: International Ovarian Tumour Analysis Assessment of Different NEoplasias in the adneXa implementation in primary care/community settings, artificial intelligence-enabled quality assurance, reconfiguration of referral pathways in primary care to reduce unnecessary referrals in younger women and consequent harm are important research areas. Systematic symptom elicitation capitalising on routine health interactions to reach underserved communities warrants further research.
Funding: This synopsis presents independent research funded by the National Institute for Health and Care Research (NIHR) Health Technology Assessment programme as award number 13/13/01.
Keywords: ADNEX; COST–CONSEQUENCE; EARLY DIAGNOSIS; HE4; INTERNATIONAL OVARIAN TUMOUR ANALYSIS; IOTA; IOTA SIMPLE RULES; ORADS; OVARIAN CANCER; RISK OF MALIGNANCY INDEX; RMI; ROMA; SYMPTOM-TRIGGERED TESTING.
When women go to their doctor with symptoms like bloating or new tummy pain or feeling full after meals or changes in toilet habits, they need tests to find out what is wrong. A tiny proportion of women with these symptoms will have ovarian cancer. We know that when caught at Stage I, more than 9 out of 10 women will survive the cancer, whereas this drops to 1–2 women when diagnosed at Stage IV cancer. Diagnosing ovarian cancer earlier will save lives and reduce the complexity and side-effects of treatment. For many years, doctors have used a test called Risk of Malignancy Index, which combines blood tests and simple ultrasound scans to determine the risk of a person having ovarian cancer. We wanted to find out if newer tests might work better and understand how much these different tests cost the National Health Service. Our research had several parts. First, we looked at all the previous studies about these tests, including information from over 30,000 women. Then, we did our own large study called ROCkeTS, where we tested 2453 women who attended hospital for the first time with these symptoms in 23 hospitals across the United Kingdom between 2015 and 2023. Each woman in our study had a blood test, a detailed ultrasound scan conducted by staff who had undergone specific training, and filled in a questionnaire about their symptoms. We collected data on whether participants underwent surgery or a biopsy within 3 months of joining the study. If they did not need surgery or biopsy, we followed them up for at least a year to make sure they were well. The old test (Risk of Malignancy Index) was not very good at finding cancer in younger women who have not been through the menopause – it missed more than half of the cancers. A newer test called International Ovarian Tumour Analysis Assessment of Different NEoplasias in the adneXa was much better at finding cancer, though it sometimes suggested women might have cancer when they did not. We found big differences in the number of women diagnosed with cancer between younger and older women. About 3 in 100 younger women and 18 in 100 older women referred to hospital for further checks were diagnosed with ovarian cancer. The new Assessment of Different NEoplasias in the adneXa ultrasound test is more detailed than current ultrasound tests and requires special training. It is like learning a new language – staff need to learn exactly how to describe what they see on the ultrasound in a very specific way. This means all sonographers (the non-medical specialists who do ultrasound scans) need specific training and regular checks to make sure they are doing the test correctly. In our study, we found that when staff were trained, the Assessment of Different NEoplasias in the adneXa test could find 9 out of 10 cancers. This is a big improvement that could help save lives. The International Ovarian Tumour Analysis Assessment of Different NEoplasias in the adneXa test also flagged up more than twenty women out of a hundred as potentially having cancer when they actually did not. We also looked carefully at how women felt about having these tests. Many women felt very anxious while having tests, especially younger women. Even when tests showed nothing was wrong, many women still worried about cancer a year later. However, when cancer was found because women had been referred by their general practitioners for suspicion of cancer, it was often caught at an early stage when treatment works better – about a quarter of the most aggressive type of ovarian cancers were found early. This suggests that referring women with symptoms for testing helps them get diagnosed early and have more successful treatment. When we looked at costs, we found that while the new Assessment of Different NEoplasias in the adneXa test was more expensive than the old Risk of Malignancy Index test, it reduced the chances of death from cancer. Modifying the scan technique by using a two-step strategy with the Assessment of Different NEoplasias in the adneXa test proved the best trade-off between cost and reducing death from cancer. A two-step strategy is where staff first look at ultrasound pictures of the growth. If it has certain innocent features (like being a single smooth-walled cyst), they can immediately tell it is almost certainly harmless – this works for about 3–4 out of every 10 cases. For the other cases that are not so obvious, staff can use the International Ovarian Tumour Analysis Assessment of Different NEoplasias in the adneXa test that looks at more details and gives a risk score. This two-step approach is performed at the same ultrasound examination and makes the process faster and simpler when possible, but still very accurate for all patients, making this approach very suitable for use in the National Health Service. The challenge now is to roll out this better test across the National Health Service. This means training sonographers in the new technique, setting up systems to check the quality of their work regularly, making sure that doctors know how to interpret the results of the scan and make decisions based on the Assessment of Different NEoplasias in the adneXa test, supporting staff to learn and use the new system confidently and finding ways to reduce waiting times for scans. Based on our research, we recommend that hospitals should switch to using the new Assessment of Different NEoplasias in the adneXa test using the two-step approach instead of the old Risk of Malignancy Index test. While this will take time and money to implement properly, it could make a real difference to women’s lives by finding cancer earlier when it is easier to treat. We also need to find better ways to support women who are worried while having tests, and research into whether International Ovarian Tumour Analysis Assessment of Different NEoplasias in the adneXa ultrasound can be used in primary care to identify women with ovarian cancer earlier and to reduce the number of women (particularly younger women) referred to hospital for more checks. An exciting area of future research is developing computer programs using artificial intelligence to help with ultrasound scans. These artificial intelligence tools could help reduce the long waiting times for ultrasound tests in the community and make sure everyone gets the same high-quality scan wherever they live. Our research has also found that when women present to their general practitioners with symptoms and they are then investigated and referred to hospital using fast track pathways, their ovarian cancer can be diagnosed early with better chances of treatment. Asking people about their symptoms during regular doctor visits may be a good way to reach underserved communities. Scientists should study this approach to see if it helps. In summary, our research has helped doctors choose the best test for women who might have ovarian cancer, which should help find cancer earlier when it is easier to treat, while making the best use of National Health Service resources.