Limitations of The Lancet papers espousing the cancer risk reduction of daily aspirin
March 21, 2012 By Dr. Kausik Ray from St. George’s University in London
Here are my thoughts on The recent Lancet papers that uses meta-analysis to study whether daily aspiring reduces the risk of cancers. I think the editorial was very good and I concur with the conclusions. There were methodological limitations:
1) The meta analyses conducted included data on about 36 000 people and about 360,000 person years of follow-up (FU). However, they excluded the two largest trials which had cancer incidence as a prespecified endpoint (the Women’s Health study and the Physicians Health study). We included those in our paper and had data on 102,000 people and 700,000 person years of FU.
2) That being said, the present data provide mechanistic insights into the potential role of aspirin. The fact that benefit on cancer incidence with low dose appears after 3 years fits with what we understand about cancer neogensis and the fact that in the second paper there were favorable effects on cancer metastases provide supportive data for a potential role for aspirin.
3) The data do not tell us much about who to treat or the risk benefits, as you need to treat about 189 people for 10 years or 1890 people for 1 year to prevent 1 new cancer. In contrast for every 219 people treated for 5 years there’s one extra major non cerebral bleed, or 1095 people/year. So the risk benefit, on average, is 2 extra bleeds per incident cancer prevented. Remember these do not capture intracerebral bleeds, retinal bleeds hospital admissions etc as cancer incidence is collected in cancer registry and is more reliable than collecting bleeding data after trial follow up has finished. The data that bleeding declines and aspirin is protective against bleeds >5 years is biologically implausible. The implication is if you treat whole populations and more bleed especially early they come off drug so the NNT (i.e. those who stay on drug) to prevent one event will become even higher.
4) The studies published add considerably to our understanding, but guidelines should not change on the basis of these for general populations. We need the results of ongoing studies which accurately record cancer incidence and document non-trivial bleeds over extended follow up to determine net benefit. Beyond that we need at an individual level to determine who is at high absolute risk of bleeding and who is at high absolute risk of cancer to determine on a personalized level whether the treatment is suitable. (e.g. those with a high bleeding risk score versus a low risk score for cancer would not be individuals the medical community should perhaps offer this treatment to.)
Ongoing trials will help clarify matters, and what the present study adds is the possibility that in a few years time, if we correctly identify risk versus benefit, we might use aspirin for the primary prevention of cancer rather than vascular disease.
The following is or previous interview with Dr. Ray on the bleeding risks of aspirin.
Biostatistics, Cardiology/medicine, Late Breaking Data, Neurology, Oncology, Preventive Medicine, St George's Univ London, The Lancet | apples49 | 
March 21, 2012 10:27 am
No Comments
No comments yet.
Notice: comments_rss_link is deprecated since version 2.5! Use post_comments_feed_link() instead. in /data/30/1/121/80/1121895/user/1185656/htdocs/wordpress/wp-includes/functions.php on line 3406
RSS feed for comments on this post.
TrackBack URI
