Effects of aspirin on cancer outcomes
Prof Peter Rothwell - University of Oxford, Oxford, UK
Interviewed by Prof Peter Elwood, University of Cardiff, Cardiff UK
Well now, Peter, you’re a neurologist; how did you get into work on aspirin and cancer?
As a neurologist my main interest is in stroke and so we prescribe a lot of aspirin, in fact my team at the moment has probably prescribed aspirin to about 20,000 patients over the last fifteen years so it’s something we use on a day to day basis. And also my two mentors, research-wise, were Charles Warlow up in Edinburgh who did one of the first randomised trials of aspirin in prevention of stroke in the UK, the UK TIA trial, and Henry Barnett, who you may know, a neurologist in North America who did the first randomised trial of aspirin in prevention of stroke full stop, the big Canadian trial. So I think I’ve always had an awareness of that work and the hint that aspirin was doing something to non-vascular death in those trials as well as to vascular events.
What gave you the brilliant idea of going back to the early vascular trials of aspirin?
Well it was partly an idea that the American aspirin trialists had already partly done in the sense that they’d followed up the Physicians’ Health study and the Women’s Health study for about ten years and found no effect on the risk of incident cancer out to ten years. So the only possibility was that there might be an effect delayed with a latency of longer than ten years. We were in a fortunate position in some ways in that we were involved in following up some of the trials that had started earlier than the American trials in the UK, Charles Warlow’s UK TI aspirin trial and Richard Dole and Richard Peto’s British Doctors’ aspirin trial. So they already potentially had twenty years’ follow up so we were able to complete that follow up and find that in fact there was something going on beyond ten years which the American trials hadn’t yet got the length of follow up that they could pick up.
Peter, in following up the subjects of the early vascular trials over up to twenty years, how did you get data on outcome? I think we’re all finding difficulties from the Office of National Statistics and other bodies in following up patients without a prior consent prior to the study being done. How did you cope?
We had some difficulties at first, yes. It took us about two or three years to get all the relevant permissions to follow up the UK TIA trial, for example. When we initially applied for local ethics approval, which was the first stage, the suggestion of the ethics committee was that we had to contact all of the surviving relatives of the trial participants to try and get permission to find out what their relatives had died from. Eventually we persuaded them that that probably wasn’t reasonable and it wasn’t possible to do that. Then we had to apply to what was called PIAG at that point, which was a national data protection committee, to get approval for anonymised record linkage. Then once we got through that committee we had to apply to another committee which ONS ran in order to repeat that process and get permission from ONS. So the whole thing took about three years. One unfortunate consequence of that was that we got the results finally showing the long-term effect on cancer about three days after Richard Dole died so he just missed that result.
You did not include the US trials, the Physicians’ Health study and the Women’s trial, you did not include those perhaps on the grounds that they were aspirin on alternate days, not daily aspirin. What do you feel about that and do you think they should be included now?
We didn’t include the alternate day trials mainly because when we first started doing this, we published the first result in 2007, the Women’s Health study only had ten year follow-up at that point, ten to twelve years, and they’d just published their own ten year follow-up. We discussed collaboration with Nancy Cook and the Women’s Health study team. I think entirely reasonably they wanted to follow up the study themselves and report their own data because the Women’s Health study was the first trial where aspirin had actually been a pre-specified outcome so I think it was very sensible that they wanted, rather than just putting their data in with everyone else’s they wanted to follow up and complete their study. I think the fact that they did that and they did that independently and they subsequently found the same result as us makes that much stronger. And we’re currently now working with Michael Gaziano and the Physicians’ Health study to do the same thing and in fact we find the same result in the Physicians’ Health study that again confirms the effect on colorectal cancer. So I think at the end of the day it’s probably worked out well in the sense that we now have these independent validations of the initial finding.
How do you yourself regard the evidence? I think we’re all convinced on gastric cancer, on colon cancer, but what about the less common cancers and what about the blood cancers which you pick out in one of your publications?
I think, as you say, there’s good evidence. There are short-term effects, as you alluded to earlier in your talk on cancer growth and metastasis which are fairly convincing but they only tell us what happens to the behaviour of cancers that are diagnosed or develop in someone already taking aspirin. So we know that if you’re taking aspirin and a cancer develops it’s less likely to be metastatic and probably grows more slowly but I don’t think that tells us very much at all about what the effect of starting aspirin after diagnosis would be which is why Add Aspirin and the other trials of aspirin in treatment of cancer are vital and are looking at a different question than we looked at. In terms of the long-term effects on prevention of cancer the colorectal cancer effect is now convincing, it’s present in five or six separate independent trials with a similar latency in each trial and a similar effect size in each trial. The data for low dose aspirin are fairly convincing now also for oesophageal cancer and stomach cancer and so there seem to be good data for cancers of the gastrointestinal lumen. The data on other cancers are still less convincing and we’re hoping by putting all the data together from all the existing trials we might make things a bit clearer but I don’t think there will be any startling effects on non-GI cancers which people haven’t found yet.
I think we still need to do some work to work out which subsets of the population might have most to gain. The Women’s Health study didn’t show any overall effect or showed a very small overall effect on long-term cancer incidence in women from, admittedly, alternate day low dose aspirin and I think we need more data on the effects of daily low dose aspirin in women. But if the effect of aspirin is confined to GI lumen cancers then they only represent about 8% of all cancers in women so even if the risk of GI lumen cancers was reduced by half by taking aspirin the impact of that on the overall risk of cancer in women would be relatively small, about a 5% reduction in the total incidence of cancer because it’s so dominated by breast cancer and gynaecological cancers. The GI lumen cancers account for about a quarter of all cancers in men and so the size of the benefit in men would be greater. I think at the moment the data suggests that the long-term cancer benefit of low dose aspirin in men is probably worthwhile, particularly if you have increased vascular risk as well it certainly adds to the benefits of taking aspirin.
And some of us here are over 50, what is your recommendation for older people with regard to aspirin prophylaxis?
Several people have looked at this and I know you’ve been involved and there have been several consensus statements that have come out which tend to support the idea that if you’re late 40s or early 50s taking aspirin for five to ten years will probably do more good than harm, will reduce the risk of cancer. If you’re at increased vascular risk it will reduce the risk of MI and possibly stroke. Given the fact that a lot of the guidelines already recommend aspirin in primary prevention in healthy men at increased vascular risk I think the cancer story just adds to the likelihood of benefit there.
How important do you think aspirin is in public health? What do you think, should it be promoted by health authorities and others?
I think we probably need to simplify the message a bit. At the moment it’s a complicated message, there are benefits from taking aspirin, there are harms from taking aspirin, the relative risks are different and the absolute risks are different and these risks differ between different subgroups of the population, they differ at different ages. So probably what we need to do as researchers is come up with some relatively simple overall measures of the balance of risk and benefit. I think the one that people understand most is just overall survival – will you live longer if you take aspirin and will you live longer in a healthy state if you take aspirin? There certainly is evidence now, putting all the data together, that taking low dose aspirin in 40s and 50s for 5-10 years will prolong life and will reduce the risk of disability. So making that message clearer and simpler is probably what we need to do.
