December 17, 2015

expert reaction to ovarian cancer screening trial

A randomised control trial of ovarian cancer screening in post-menopausal women was published in the Lancet. Screening involved annually testing blood for levels of the protein CA125, and was compared with transvaginal ultrasound screening and with no screening. The study reported that blood screening may result in a reduction in deaths from ovarian cancer but it also gave false positives, and that further research is needed to further investigate the size of any reduction in mortality and the cost effectiveness of any proposed screening programme.

Prof. Kevin McConway, Professor of Applied Statistics, The Open University, said:

“The problem with easily making sense of this study is that there are several different statistical analyses (summarised in Table 3 in the paper), and they all gave slightly different confidence intervals (a way of indicating how big or small the true reduction in mortality from screening might plausibly be). The primary planned analysis on their primary outcome (the Cox model on ovarian cancer mortality; at the top of table 3) showed reductions in deaths where the central estimate was reasonably substantial but the confidence intervals included zero, so that was not a statistically significant result and in the usual statistical terms could plausibly be due to random variation only. But because of various features of how the data looked, the authors did other analyses (the Royston-Parmar model), which gave results that are broadly similar (reduction in deaths on the new screening method, MMS, compared to no screening of 15 or 16 per cent, with confidence intervals going from a little bit below 0% to somewhere round 30%). In addition, though, they did some other statistical analyses – splitting up the follow-up times in the Royston-Parmar model to 0-7 years and 7-14 years instead of looking at all the years together, using a different statistical test (the weighted logrank or WLR test), and omitting the women who had undiagnosed ovarian cancer when they first joined the study, and some of those did give larger reductions in deaths, two of which were statistically significant (just).

“Now, doing these extra analyses can be seen as an appropriate post hoc response to how the data turned out to look (which in some respects weren’t as they originally expected), and the authors give justifications for what they did. But equally it is also the case that the more analyses done, the more likely it is that one of the results will come out as positive. I think the author’s overall conclusion, that these results show promise but they need a few more years’ follow-up to know exactly what’s going on, may well be reasonable. But their last sentence is important: “Further follow-up is needed to assess the extent of the mortality reduction before firm conclusions can be reached on the long-term efficacy and cost-effectiveness of ovarian cancer screening.” That further follow-up might possibly lead to the conclusion that, on balance, screening isn’t worth it. The results are promising, but perhaps not all that promising. The MMS screening missed 41% of the ovarian cancer cases. The authors report that the number of false positives who were operated on was “fewer than the upper acceptable limit” for MMS, but there were still quite a few, and the complication rate in operations on false positives is about 3%, which isn’t huge but needs to be take into consideration.”

Prof. Sir David Spiegelhalter, Winton Professor of the Public Understanding of Risk, University of Cambridge, said:

“The researchers unfortunately initially chose an inappropriate primary analysis.  First, they assumed the intervention had a constant relative effect throughout all follow-up, whereas there is considerable delay before such screening might show benefit.  Second, they included cases who turned out already to have ovarian cancer.

“The plots show clearly that there is no effect for the first 7 or so years, but after that there is some evidence of a benefit, particularly if the cases who already had ovarian cancer at screening were left out.

“Of course this does not mean that a screening programme is necessarily appropriate – the cost-effectiveness would need to be carefully assessed taking into account long-term follow-up data.”

Dr Clare Mckenzie, consultant gynaecological oncologist and Vice President for the Royal College of Obstetricians and Gynaecologists (RCOG) said:

“Although still at an early stage, the initial results from the UKCTOCS trial into screening for ovarian cancer are promising. This large scale, randomised control trial followed 200,000 post-menopausal women for 14 years at 13 centres across the UK. The study consisted of annually testing blood for levels of the protein CA125, and was compared to yearly screening using ultrasound and no screening.

“During that time 1,282 women were diagnosed with ovarian cancer. Results so far suggest that approximately 15 deaths could be prevented by CA125 screening for every 10,000 women screened. However for every woman with a positive screen who underwent surgery and was found to have ovarian cancer, two did not.

“This study is important in that the early detection of ovarian cancer, and hence early treatment, has the potential to save lives. However, longer follow up is needed to determine how effective the test is. Women who are worried about ovarian cancer should talk to their doctor who can explain their risk of cancer and available tests.”

Prof. Christina Fotopoulou, an ovarian cancer surgeon from the Ovarian Cancer Action Research Centre, said:

“The latest UKCTOCs findings are promising and a welcome step towards a screening method for ovarian cancer – a disease that takes the lives of more than 4000 women in the UK each year.

“We’ve always been one step behind the disease but a screening method would allow us to catch it at a stage that makes treatment more effective. While we’re not there yet, and will need to wait for further study and definitive results, UKCTOCS shows that research in the field is gaining pace.

“In the meantime, it’s important that we keep dedicating time and resources to researching ovarian cancer, to find a way to improve survival rates.”

Dr Adam Shaw, Clinical Lead for Cancer Genetics, Guy’s & St Thomas’ NHS Foundation Trust, said:

“The potential this study shows for earlier detection of ovarian cancer and the number of lives saved is very encouraging. However, false positives meant that three women had surgery for every one case of ovarian cancer diagnosed. Although this may be acceptable in terms of mortality benefit and comparison with other cancer screening programmes, there are significant clinical and resource implications of this screening method.

“Further work to assess patient acceptability, the proportion of women who are likely engage with screening across different communities, and a health economics evaluation will be important future steps. Only then will we know whether patients feel the level of false positives is acceptable and whether survival benefits justify the complications and costs of unnecessary surgery as well as the burden of population screening.

“Nonetheless, this study is a landmark step in devising effective screening for ovarian cancer, which is often portrayed as the silent killer.”

Prof. Shirley Hodgson, Professor of Cancer Genetics, St George’s, University of London, said:

“This paper is from an experienced group of researchers who have an excellent reputation in studies evaluating screening for ovarian cancer. This research evaluated multimodal screening in comparison with ultrasound alone or no screening in a very large cohort of postmenopausal women not at genetically increased risk for ovarian cancer. Multimodal screening (MMS) involved testing of blood levels of ca125 (a serum marker that goes up in ovarian cancer but is rather non-specific and variable and changes over time), using a protocol that determined whether further ca125 levels should be monitored every three months and whether ultrasound screening should be added. The end point for analysis was death from ovarian cancer and an additional endpoint was death from peritoneal cancer (allied to ovarian cancer).

“The outcome was that there was a reduction in ovarian cancer mortality in the group screened by MMS when compared with the other groups, and this effect was significantly higher after 7 years f follow up. There was a smaller reduction seen in the group screened by ultrasound alone.

“This is an important finding and should be followed up with further studies of this kind to confirm the results before considering whether MMS should be implemented in the general population.”

‘Ovarian cancer screening and mortality in the UK Collaborative Trial of Ovarian Cancer Screening (UKCTOCS): a randomised controlled trial’ by Ian J Jacobs et al. published in the Lancet on Thursday 17 December 2015. 

Declared interests

Prof. Kevin McConway, Prof. Sir David Spiegelhalter, Dr Clare Mckenzie and Prof. Christina Fotopoulou declare that they have no relevant interests.

Dr Adam Shaw: “I declare that I have no personal, professional or financial interests in the outcomes of this study.”

No others received