Subscribe
Share
Share
Embed
The Trial Talk Podcast explores how our work at the MRC Clinical Trials Unit at UCL is improving health in the UK and worldwide. In this new series, we will hear from world-leading experts about the studies we carry out. We will get inside trials on cancer, infections, and neurodegenerative diseases, explore how public and patient involvement is shaping our studies, and discover new ways to run smarter studies.
Lessons from UKCTOCS, a large-scale trial in ovarian cancer screening
Summary Keywords:
MRC CTU at UCL, ovarian cancer, screening trial, UKCTOCS
Speakers:
Guest: Usha Menon
Host: Berta Terre-Torras
Transcript:
00:00:00:00 - 00:00:41:13
Berta Terre-Torras
Hello and welcome to the Trial Talk podcast. I'm your host, Berta Terre-Torras, a former researcher and a science communications officer at the MRC Clinical Trials Unit at UCL. In this show, we explore the clinical trial landscape by talking to the clinicians, researchers and patients that are behind the work we do. If you're interested in learning how our research can help improve healthcare in the UK and around the world…
00:00:41:15 - 00:01:16:15
Berta Terre-Torras
This is the podcast for you. This year is a special one for the Unit as we are celebrating our 25th anniversary. That's 25 years of smarter studies, global impact and better health. In 2001, we began running one of the world's largest trials, the UK Collaborative Trial of Ovarian Cancer Screening or UKCTOCS. We run it in partnership with the Institute for Women's Health at UCL.
00:01:16:17 - 00:01:52:13
Berta Terre-Torras
The trial looked at different ways of detecting ovarian cancer earlier, in more than 202,000 women. Unfortunately, the approaches tested in UKCTOCS didn't save lives. We have a whole episode explaining the long-term results of the trial. You can listen to that episode via a link in the description. In today's episode, we will discuss the main challenges that the UKCTOCS trial faced in recruitment and various aspects of design and analysis, and how the team addressed them.
00:01:52:15 - 00:02:24:18
Berta Terre-Torras
Although the trial was designed in the late 90s, many of the approaches and challenges are still relevant today for many trials, particularly screening trials. We will talk about the lessons learned from conducting such a large trial, which are published on the NIHR website. For that, I will be speaking to Usha Menon, professor of Gynecological Oncology at the MRC CTU at UCL, and one of the chief investigators of UK C Talks.
00:02:24:20 - 00:02:36:19
Usha Menon
I'm Usha Menin and I co-led the trial in the first decade with Professor Jacobs and then continued as chief investigator during the extended follow-up phase.
00:02:36:21 - 00:02:54:09
Berta Terre-Torras
When planning a clinical trial, it's important to identify one main research question that the trial will try to answer. This is also called the primary outcome. So, what was the main research question for UKCTOCS?
00:02:54:11 - 00:03:32:16
Usha Menon
So the main result, we were trying to find out whether screening would decrease the number of deaths from ovarian cancer in the screen arm compared to the no screening arm, which we call the control arm. This primary outcome determines how big the trial needs to be and usually we specify it at the start of the trial. We were looking to decrease deaths by 35% between the control arm and the screen arm.
00:03:32:18 - 00:03:41:11
Berta Terre-Torras
And how did you decide how many people needed to take part to ensure that the trial would provide meaningful results?
00:03:41:13 - 00:04:12:04
Usha Menon
That is called the sample size calculation in trials. We tried to estimate how many participants need to take part in the trial and how long the trial needs to be. For that, what we do is we determined statistically what numbers would give us the power, and by that I mean we could specify 80% or 90% surety that whatever result we got was true.
00:04:12:06 - 00:04:44:01
Usha Menon
To do that in UKCTOCS, we had to get a certain number of deaths in the control and the no screening arm, and then in the screening arm, the deaths needed to be 35% less. So, if there were 100 deaths in the control arm, it needed to be 65% in the screen arm. And then we had to make sure that that number was big enough for us to be certain that it was a true result and not a random result.
00:04:44:03 - 00:05:13:03
Usha Menon
So the key as you can understand, is the control arm or the no screening. We specified beforehand the number of deaths we need to get in that. In UKCTOCS that number was 222. When 220 women had died of ovarian cancer in the control arm. We could then look at the study arm to see where the 35% less had died in the study arm.
00:05:13:05 - 00:05:25:13
Berta Terre-Torras
And how d to estimate the total number of participants that you needed to make sure that 222 women would die in the control arm during the course of the trial?
00:05:25:15 - 00:05:51:10
Usha Menon
We used the national statistics because we know how many women die of ovarian cancer every year. At that point when we started UKCTOCS, which is 2000, that number was 37 ovarian cancer deaths, 100,000 women. So using that, we thought there would be 37 deaths every year. So one could calculate you get 222 deaths. How many years it would take in the control arm.
00:05:51:12 - 00:06:18:05
Usha Menon
And that was how we determined how many years we would follow up, how many women we would have in the trial. But we ran into two problems. These are problems which are very common in trials. So it is not just our trial. What we did was we just took the national statistics and then used that to estimate when we would reach to 222 deaths in the control arm.
00:06:18:07 - 00:06:59:08
Usha Menon
But in reality we had something called the eligibility criteria effect. And what I mean by that is only the women who did not have ovarian cancer were allowed to join the trial. And really if you think about it carefully, most women die from ovarian cancer about five years after they are diagnosed with the disease. So as you can imagine, in the first two or three years, we had very few deaths from ovarian cancer in the trial because only included in the trial were women who developed ovarian cancer after they had joined the trial.
00:06:59:10 - 00:07:26:24
Usha Menon
So that was the first issue. When we were five years down the line, we could see in the control arm that there were not enough deaths as we had predicted there would be in the no screen arm. So if we had stopped the trial as we had thought it, at seven years from randomization, your participation in the trial is over, and that is how we had conceived the trial.
00:07:27:01 - 00:08:09:03
Usha Menon
But it would be that if we had done that, we would not have got the power to get a true result because the number of deaths that needed to occur in the no screening arm or control arm would not have happened. The other issue that we came across and which we had not anticipated, but is that when we invite people to take part in a screening trial or a prevention trial, it is those women who are interested in health, in screening interventions, in contributing to the let's say, better health decisions to take part in the trial.
00:08:09:05 - 00:08:33:03
Usha Menon
So they are inherently somewhat different from the whole population, and they probably do many more healthy things. They probably don't smoke, they eat more healthy food, they might do more exercise. And we would see that again about five, six years down the line, when we looked at the no screening arm. Where we had done nothing, the women had just joined the trial.
00:08:33:08 - 00:09:00:23
Usha Menon
You could see that when we compare to the national statistics, deaths were down from cardiovascular disease, from respiratory disease. Many cancer incidents were down, heart attack rate was lower. And this we called the healthy volunteer effect and we did not take that into account. And so, again, in the control arm that was a second reason why events were lower.
00:09:01:00 - 00:09:08:04
Usha Menon
For both these reasons, we had to then continue the trial for more years than be thought initially.
00:09:08:06 - 00:09:20:18
Berta Terre-Torras
Now that you are aware of how the eligibility criteria effect and the healthy volunteer effect can influence your predictions of the size of your trial, what would you do differently?
00:09:20:20 - 00:09:51:11
Usha Menon
I would suggest that it's really important, most of us always do look at national statistics to estimate events in the control arm. But it's important to add a factor for the healthy volunteer effect. And if you did have an eligibility criteria that might impact on your primary outcome, the key result that you need to also consider that and make some calculations for that.
00:09:51:13 - 00:10:03:13
Berta Terre-Torras
In UKCTOCS, however, you encounter these two problems once you are a few years into the trial. So how did you overcome them whilst the trial was still running?
00:10:03:15 - 00:10:34:14
Usha Menon
The main issue was that in the control arm, the only way we could reach that target of 222 deaths from ovarian cancer, which we could look to see if we had saved lives in the screen arm. To reach that, it was only possible in two ways One is recruiting more women into the trial, and two is by waiting for longer because each year some women died from ovarian cancer, unfortunately.
00:10:34:14 - 00:11:12:15
Usha Menon
Because we had already finished the recruitment, which would have been very hard for the centers to again restart recruitment, recruitment had been finished two, three years prior. So we decided that we would continue the trial, continue to follow up of the control arm till the time when the number of events that had to happen occurred. But then there was an issue: what would we do with the screen arm? Now if we stopped the screening arm and the women didn't have screening, then of course the screen arm would look very much like the control arm without screening.
00:11:12:15 - 00:11:42:16
Usha Menon
So there would be what we call a dilution effect. So we continued screening. So finally, what we did was, the women were recruited between 2001 and 2005. Instead of six years of screening, we asked each woman if she would continue attending for screening every year till December 2011. So this meant that some women had seven years of screening and some women had 11 years of screening, depending on which year they joined the trial.
00:11:42:18 - 00:12:06:22
Usha Menon
And then we followed up for three more years. So just to say that if you are planning the trial, you can now look at the healthy volunteer effect in different sorts of cohorts like in our study, you can look at the UK Biobank cohort where they invited a lot of people and then you can see the people who joined the trial and look at their mortality data.
00:12:07:01 - 00:12:32:04
Usha Menon
So this was really very much more in a screening or prevention trial. Because in a treatment trial, if I had cancer and you offered me a trial, it is most likely more people take part. It's just that in these big population trials, usually like it's somewhere between 10 and 20% of those who are invited who take part.
00:12:32:06 - 00:13:05:11
Berta Terre-Torras
Another issue that UKCTOCS faced was to do with the analysis of deaths between the screening arm and the no screening arm. The original statistical analysis was defined in 1999 when the trial started. During all the years that the trial was running, the team had more data available on screening trials and ovarian cancer mortality. At that point, it was clear that the original statistical approach wasn't ideal anymore and it needed to change.
00:13:05:13 - 00:13:08:17
Berta Terre-Torras
How did you implement that change?
00:13:08:19 - 00:14:00:23
Usha Menon
We did change the approach for a statistical reason, and that is very trial specific and specific to screening trials. What we did was we had an opinion, of course, a lot of discussion with the trial management team as well as in the trial steering committee. But to make sure that we were not biased in any way, we asked 12 international experts in statistics, in clinical trials and in screening, to consider the issue that we were facing and to come up with a solution. And they sort of came to the same conclusions that we as a team had come to looking at all the evidence. And they supported us and they suggested we change the approach.
00:14:01:00 - 00:14:30:24
Usha Menon
We then published the entire dilemma and the pros and cons of changing the approach and why we change approach and the opinion of the various independent members. And I think it was that publication with the transparency about the process that made it possible to do this change. So what I want to highlight is if it's scientifically justified, you can make a change.
00:14:30:24 - 00:14:51:01
Usha Menon
What is important is that it's absolutely transparent and you try to take into account as much independent opinion as possible. Because of this process, when we finally analysed the data, there was no question about the change of approach.
00:14:51:03 - 00:15:12:09
Berta Terre-Torras
Looking now into the challenges of conducting such a large trial, one relevant aspect to consider is recruitment. UKCTOCS aimed to recruit 200,000 women, which is an enormous number of participants. How do you make sure you could recruit all these women?
00:15:12:11 - 00:15:43:06
Usha Menon
Oh, we had to recruit 200,000 women and we were perhaps one of the first trials to use electronics health records. This is now much more freely available and much better organized through NHS Digital and NHS DigiTrials and many of the current trials are looking to use these resources for recruitment. But back in 2000, we had to do a slight modification of what is available now.
00:15:43:08 - 00:16:17:13
Usha Menon
We requested and we were given personal data of the women and their GP's as well as the NHS number women age 50 to 74, who we could then mail, merge and send personal invitation letters to. And what we did was just so that women were not kept waiting for a long time. They said they would like to join a trial or we were transferred lists of 5 to 10,000 women every three months from NHS age sex registry.
00:16:17:15 - 00:16:35:01
Usha Menon
These were women living around the 13 centers who were taking part in the trial. Because we were able to send out so many invites, we were able to ensure that we had enough women to join the trial.
00:16:35:03 - 00:16:40:12
Berta Terre-Torras
And how long did it take to recruit 200,000 women?
00:16:40:14 - 00:16:47:15
Usha Menon
In reality took us five years. That was we have thought four, but it took us 4.8 years.
00:16:47:17 - 00:17:06:10
Berta Terre-Torras
Because of recruitment lasted five years. That meant that some women joined the trial in 2001, but others in 2006. How did you ensure that the first woman and the last woman that enter the trial were kept in form in the same way?
00:17:06:12 - 00:17:42:08
Usha Menon
So the thing we were committed to from the beginning was one that all the women would receive the same kind of information, whether they came to Belfast or Nottingham or London or Portsmouth. And the other thing was that whether they were recruited in 2001 or 2005, the quality of information and the informed consent process was similar. So to do that, again, you know, we undertook some measures which again are very probably widely used these days.
00:17:42:08 - 00:18:17:16
Usha Menon
At that time, you must remember there was only video, so we made a video where some of the members of our team and their parents really and aunts and grandmothers, because we wanted older women, sat around and asked questions and discussed the trial. And so we had a BBC team help us make that video and we set up monitors in all the child recruitment centers and the women would come in groups and see the video before they participated in a group discussion.
00:18:17:16 - 00:18:40:24
Usha Menon
So this was the other thing we had. We had a group discussion following the videos so that at least some of the quieter women would have an opportunity to hear the questions asked by others. And this after this, we went on to the usual 1 to 1 recruitment discussion with the nurse and the signing of the consent.
00:18:41:01 - 00:18:52:07
Berta Terre-Torras
Another element to consider when thinking about recruitment is the role of patient advocacy. How were the interactions with patient groups and charities in UKCTOCS?
00:18:52:07 - 00:19:20:23
Usha Menon
We had a lot of interaction because the whole question at trial was really driven by the patient groups. It was because of them and the ovarian cancer charities, that we had extra leverage. When we put in the grant application, there was a real desire from them that we should have a trial to see if anything could be done to decrease the mortality from ovarian cancer.
00:19:21:00 - 00:19:46:24
Usha Menon
So we had members from these patient groups sitting on various committees having both oversight as well as contributing to the discussions. And they looked at a patient information material. But I do feel that if we did it again now, we might have had even more interaction.
00:19:47:01 - 00:19:53:17
Berta Terre-Torras
And were there any other key elements that help such a big trial run smoothly?
00:19:53:19 - 00:20:19:20
Usha Menon
I think in any trial you have to be very proactive. The thing we did was we built a trial management system, which was more than just an EDC or an electronic data capture system. It, of course captured data, but it had a lot of functionality and it was automated and it would do many things on its own so that things were done timely.
00:20:19:22 - 00:20:53:12
Usha Menon
So, you know, it was like a record of everything of the parties, as I said, you know, used to get these lists from the various primary care trusts, from the NHS registers, the lists of women, and we would just upload these lists into the trial management system. The system would then print, sent out letters. And as soon as you ticked that the lady had responded, it would automatically schedule appointments, send out letters to those women, invite them.
00:20:53:14 - 00:21:19:24
Usha Menon
And what we did was to make sure that the recruitment clinics were full and we were meeting monthly targets. We would vary the recruitment. So we were every week monitoring recruitment at the various centers and forecasting what it might be. Because one of the things was, of course, women accepting the invitation, the other was women turning up for recruitment.
00:21:20:01 - 00:22:00:19
Usha Menon
But in addition, what we did was it was clear that some centers were struggling. So we had what we called blitz clinics, and that was all hands on board. The entire central team at UCL, the UKCTOCS team, all of us would go down to that center and the consultant at that center, that is the PI at that centre, other additional kind of fellows at that center, the team of the center, all of us on a Saturday and Sunday would do an open recruitment clinic where we would try to catch, make up and recruit 200 women in the course of two days.
00:22:00:19 - 00:22:30:01
Usha Menon
We never hit 200, but maybe about 150. It also created a very good sense of camaraderie. It was clear that we were one team with one goal as opposed to a Nottingham team, Manchester team, a Belfast team. So, you know, we shared our problems and we did help each other. We tried to make sure that there were no lagging centers, that everyone was moving together.
00:22:30:03 - 00:22:42:22
Berta Terre-Torras
Once the recruitment of 200,000 women was completed, how did you ensure that all these women stayed in the trial during the following years.
00:22:42:24 - 00:23:14:05
Usha Menon
Followed attention. One of the things was all women were older and it was very early in the Internet data. So I would say perhaps only one sixth of them had email addresses. We didn't even try to use email addresses, which would be the defining now. For us, the retention was most important. One of the control women would anyway sign that we could follow them up through the NHS registries.
00:23:14:07 - 00:23:41:19
Usha Menon
But in the screen cohort you do ensure that they came year on year for screening and to do that we give them as much flexibility as we could with regards to changing their appointments. They could ring and reschedule appointments because of the trial management system. So everything was like on our trial management system. So I could look up somebody's appointment at Belfast and they asked me to change it.
00:23:41:19 - 00:24:09:01
Usha Menon
I could look for another appointment and on the phone from London I could change the appointment. The rule was if a phone rang, whoever you are, you must pick it up, because mostly it was participants ringing and we were very clear that the whole of compliance was based on interaction and how we behaved and how welcome we made the participants when they came for screening.
00:24:09:03 - 00:24:33:18
Usha Menon
And I must say we had great teams at the centers. They knew many of the women by name. So each of the centers finally had about 15-20,000 women, of whom 10,000 were coming for screening every year. But within a couple of years, many of the, we had the nursing staff and the receptionist who stayed with us quite often throughout the trial.
00:24:33:18 - 00:24:52:03
Usha Menon
So they knew some of the women, many of the women by name, and they would check with them in and help them as best they can. So it was mainly because of the interaction between us and the participants that we had very high retention right until the end of the trial.
00:24:52:05 - 00:25:16:11
Berta Terre-Torras
The UKCTOCS trial ended in 2020 and the main results were published the following year. Since then, the team has been working to explore and manage a huge collection of samples and data that were obtained during the trial. This are an invaluable resource to help advance our understanding of ovarian cancer.
00:25:16:15 - 00:26:06:02
Usha Menon
You know, as during the course of UKCTOCS, we had a biobank generated of data and samples donated by the women that could be used for secondary research. Looking for new tests, understanding the natural history of disease. And this is not limited just to ovarian cancer because these 200,000 women developed all sorts of diseases and cancers in the course of the 16 years that we followed them up. And so while we are most interested in finding a better test for ovarian cancer using these samples, there are groups we work with who are working on pancreatic cancer, on breast cancer, on colon cancer.
00:26:06:04 - 00:26:12:07
Usha Menon
So it's a very rich source for early detection.
00:26:12:09 - 00:26:59:05
Berta Terre-Torras
And this is the end of today's episode. Thank you very much for listening. If you want to find out more about the insights from running UKCTOCS, you can read the full article on the NIHR website. You will find a link in the description. Or you can visit our UKCTOCS website at ukctocs.mrcctu.ucl.ac.uk. And don't forget to follow us on Twitter at MRC CTU and on LinkedIn at MRC Clinical Trials Unit at UCL for the latest updates. See you in the next episode of the Trial Talk.