* Low dose aspirin has been the subject of much research in the fields of cancer and cardiovascular disease. It has been studied for potential prevention and help post diagnosis.
* This week’s note examines a very large systematic review and meta-analysis of 118 studies, which reviewed aspirin for cancer post diagnosis.
* Significant reductions (approximately 20%) in mortality were found in patients taking aspirin for some of the most common cancers and other cancers when grouped together.
* The paper summarised some findings on extension of median survival following cancer diagnosis, which is the way in which cancer data are presented. Some of the numbers on possible life extension among patients taking aspirin were impressive.
* The mechanisms were not explored in this paper but were explored in another paper by the lead author. The mechanisms are plentiful and plausible.
* The one downside of taking aspirin is the risk of bleeding. This note reviews the evidence for the risk of bleeding and tries to put possible benefit and risk in context with each other.
* There is no evidence that aspirin will increase the risk of fatal bleeds and so the concern that possible life extension may be offset by a fatal bleed is not warranted.
* The authors concluded that the evidence should be shared with patients and their carers so that they can make an informed choice about whether to add aspirin to their cancer treatment. I agree with this conclusion.
This is the third week in a row where I’ll declare an interest. I have met four of the authors on this week’s paper. The lead author, Professor Peter Elwood, was an examiner when I defended my PhD. Professor Elwood presided over one of the most well-known epidemiological studies – the Caerphilly Cohort study, which started in 1979 (Ref 1). I thought I was doomed with my findings that there was no evidence against total or saturated fat but he had found the same at the time. We both concluded that dietary fat guidelines had been introduced without evidence.
I met Janet Pickering with Peter once when we jointly explored the Caerphilly data to see if any other dietary associations could be observed. We concluded at the time that they couldn’t. I have met John Watkins a couple of times, as he and his wife are close friends of our neighbours. I also contacted John early in the epidemiology part of my PhD for advice, as he has worked in this field all his life. I met Christine Delon at the Public Health Collaboration conference in 2019 and we have emailed each other about research a few times since.
Peter has had many interests over many years, culminating in over 700 publications with his name on. He officially retired in 1995 but he was granted an honorary chair at Cardiff University, where he still goes to work. Peter quotes Archie Cochrane (founder of the Cochrane institute and his predecessor as Director of the MRC epidemiology unit) as saying “To treat a disease is an admission of failure, Prevention is the ultimate aim of medicine.” For over 50 years Peter’s research has been focused on the prevention of disease.
Alongside dietary fat and heart disease, another of Peter’s great interests has been aspirin. The summary description issued for a talk which Peter gave in March 2018 explained “Forty years ago a randomised controlled trial of aspirin in the prevention of heart attacks showed a 25% reduction in deaths and brought aspirin to the notice of cardiologists. Ten years later benefit was confirmed in one of the first ‘overviews’ conducted in medicine. In 2011 a series of follow-up studies of randomised trials showed a 40% reduction in cancer deaths and brought aspirin to the notice of oncologists. A reduction by aspirin in deaths and in the spread of cancer has been confirmed in numerous overviews.
But, aspirin is highly controversial. Its benefits in both vascular disease and cancer are widely acknowledged, but its dangers in increasing stomach bleeds are held by many to outweigh the benefits” (Ref 2).
Peter is an enthusiastic recipient of the Monday note and he invited me to review one of his recent publications (July 2021) entitled “Aspirin and cancer survival: a systematic review and meta-analysis of 118 observational studies of aspirin and 18 cancers” (Ref 3).
The title tells us so much already – it’s a huge study. It’s the best kind of evidence – a systematic review to find all the studies and then pooling them together in meta-analysis. It’s also looking at survival – which is what I would want to know had I had a cancer diagnosis. With the bizarre sense of evaluating the evaluator – here goes.
The study opened with an explanation of the background and why this review was undertaken. "Despite the accumulation of research papers on aspirin and cancer, there is doubt as to whether or not aspirin is an acceptable and effective adjunct treatment of cancer." The debate is as the summary accompanying the talk set out above – there is evidence of benefit, but there is concern that harm may outweigh this.
Elwood et al conducted a systematic review of studies of patients with cancer some of whom took aspirin having received their cancer diagnosis. There were 118 observational studies relevant to the search. Eighty-one studies reported on aspirin and cancer mortality and 63 studies reported on all-cause mortality (some reported on both). Approximately 20-25% of patients reported taking aspirin, which totalled about a quarter of a million people.
The study found aspirin to be associated with a reduction of approximately 20% in cancer deaths (pooled hazard ratio (HR): 0.79; 95% confidence intervals: 0.73, 0.84 in 70 reports) with similar reductions in all-cause mortality (HR: 0.80; 0.74, 0.86 in 56 studies). The corresponding author of every paper was written to asking for information on bleeding. The frequency of bleeding increased in the patients taking aspirin "but fatal bleeding was rare and no author reported a significant excess in fatal bleeds associated with aspirin. No author mentioned cerebral bleeding in the patients they had followed." We’ll come back to bleeding.
Most of the 118 studies examined three of the most common cancers – colon, breast and prostate. Supplementary File 4 set out the meta-analyses of aspirin and cancer mortality for each of these three cancers separately. The first two achieved significant differences: colon (HR: 0.72; 0.63, 0.82); and breast cancer (HR: 0.84; 0.72, 0.98). The prostate cancer result was not statistically significant (HR: 0.89; 0.78, 1.02). Pooling together evidence for aspirin and mortality in other cancers (e.g., lung, ovarian, liver, bladder, melanoma etc) also achieved a significant reduction (HR: 0.79; 0.70, 0.88).
When all-cause mortality was reviewed, aspirin continued to be associated with a significant difference for those with colon cancer (HR: 0.83; 0.75, 0.92), but not for those with breast cancer or prostate cancer. There was a significant difference for those who took aspirin with other cancers for all-cause mortality (HR: 0.67; 0.60, 0.75).
The rigor of the paper was the best I’ve seen. Pages and pages in the supplemental file were devoted to various tests that can be undertaken with meta-analysis to confirm whether results are robust or if any bias has been detected. Let’s use publication bias, as an example. Publication bias is the tendency to decide to publish a study based on the results of the study, rather than on the basis of its theoretical or methodological quality. In the absence of publication bias, studies are distributed symmetrically about the effect (HR) found. There is a diagram called a funnel plot, which can show asymmetry in the presence of bias. Other tests can also be done to detect publication bias and other potential bias in the pool of research.
The researchers found that the result for colon cancer mortality was robust, but not the result for colon cancer and deaths from any cause. The breast cancer mortality result was not robust. The findings for other cancers as a group – for death from cancer and death from any cause – were both robust.
The most powerful part of the paper for me was where median survival data were presented. Cancer has always been a field where I think data are presented in the ‘right’ way and by this I mean the way in which patients can understand what they want to understand. Cancer drugs invariably have their outcomes presented in terms of median survival. If you remember the three averages – mean is the sum of all the data divided by the number of data points, mode is the most common data point and median is the middle in an ordered list. Cancer drugs are presented as – Drug X resulted in a median survival of (say) 24 months vs a placebo median survival of (say) 12 months. The patient can then decide if the side effects of Drug X are worth the predicted 12 additional months. In contrast, cardiovascular disease interventions are presented as – take this drug for 5 years and you have a 20% (relative) reduced chance of having an event, which might turn out to be a 1 in 150 absolute risk of that event and then you might have that event on day 1 of year 6. I like how cancer does data.
Elwood et al reported that a number of the papers they examined provided estimates of the association with aspirin in terms of the additional expected survival. Several examples were given (all references are in the paper, p6). Albandar et al followed 174 US veterans with colorectal cancer to death and reported that the median survival of patients taking aspirin was 941 versus 384 days in those not taking aspirin. “Several papers record an increased survival associated with aspirin taken by patients with liver cancer: in one 18 months additional survival; in another 6% more patients survived 10 years with aspirin after diagnosis, and the median overall survival period after embolisation was longer for patients taking aspirin (57 versus 23 months). In a study of endometrial cancer, 91% of patients taking aspirin survived 10 years compared with 81% of patients not on aspirin. In a study of patients with lung cancer, patients on aspirin survived 1.69 and only 1.02 years if not on aspirin. In a study of pancreatic cancer, the 3-year survival was reported to be 61% in patients taking aspirin versus 26% in patients not taking aspirin, and finally, the 3-year survival of US veterans with head and neck cancer was 79% in those taking aspirin, compared with only 56% of those not taking aspirin.”
I’ve cited much of that section verbatim as it was so powerful. Those are large differences for someone given months to live. Thanks to Peter’s work, I was aware of some of these differences as I researched them when my brother was diagnosed with cancer in May 2017. He has taken low dose aspirin daily since diagnosis and I’m delighted to say that he remains in excellent health, as fit and active as he was over four years ago. That could be nothing to do with aspirin of course.
This particular paper didn’t go into the mechanism by which aspirin can help with cancer. I asked Peter for this, and I did some research. The research I did elicited inflammation as an important plausible mechanism. The University of Texas cancer centre quoted their professor of Gastroenterology, Hepatology and Nutrition, Dr Robert Bresalier, “Aspirin reduces the risk of cancer by fighting inflammation.” Inflammation is a double edged sword – it is a key part of our immune system’s response to sickness, injury or disease “But chronic or prolonged inflammation can create an environment in which cancer thrives,” Bresalier said. He continued “Aspirin blocks the production of the enzymes that increase inflammation in your body and speed or assist the growth of cancer cells” (Ref 4).
Peter shared with me a story based on the fact that years of botanical research have shown salicylates to be highly active in plants. “Early in my setting up the first RCT of aspirin and vascular disease mortality in 1974, I visited a dear old man in a research botanical unit – Prof Stan Pierpoint, said at that time to be the world expert on salicylates. I told him about my RCT on vascular disease and he said: ‘Skip vascular disease, go for cancer!’ I had to explain to him that we had enough resources for a vascular trial with adequate power to detect a small reduction in vascular disease. Had we wanted to test aspirin in cancer we would have needed about ten times the numbers and ten times the time! But how right he was!”
Peter also shared an advance copy of another paper of his, which is awaiting decision on publication. This not-yet-published paper contains substantial evidence about the biological mechanisms relevant to cancer – both prevention and help post diagnosis. The paper discusses how aspirin can disrupt enzymes and interfere with pathways in a beneficial way. Aspirin can also inhibit other harmful pathways. The paper reports that almost 60 years ago there was evidence that aspirin could reduce metastatic spread of cancer – spread, of course, is what can make some manageable cancers fatal.
The observation in the Elwood et al ‘118 studies paper’ that aspirin is associated with a significant reduced risk of death from any cause, and not just death from cancer, could be explained by the paper undergoing peer review. It reports that "Patients with cancer appear to be in a hypercoagulable state with marked increases in vascular and thromboembolic disease events." In the field of cardiovascular disease, aspirin has been shown to have an anti-coagulant (anti-clotting) effect.
The evidence for benefit and the plausible mechanism are there. Yes, the risk ratios of approximately 20% are not Bradford Hill causation territory, but I would be interested in a possible extension of life span with an easy intervention if I had been diagnosed with cancer. The ‘but’ is the bleeding with which aspirin is also associated.
Bleeding was not extensively addressed in the paper we’re reviewing this week since “Many of the studies however had been based on recorded data, with no direct contact with the patients involved, and authors of such reports had little or no knowledge about bleeding in the patients they described.”
The paper in peer review addresses bleeding more extensively. It did what I would do – a review of systematic reviews/meta-analyses examining low dose aspirin and bleeding. It also split the review into bleeding generally and fatal bleeding, the latter, of course, being the ‘side-effect’ of most concern. However, arguably, as we are weighing up the benefit vs harm of low dose aspirin for cancer post diagnosis, we should only look at data examining bleeds in cancer post diagnosis. If we take benefit for cancer and harm for every possible circumstance in which someone takes aspirin (cardiovascular disease, post-surgery to avoid clotting, when flying to avoid DVT etc), we are not comparing like with like. Plus, there may be something related to cancer whereby aspirin increases or reduces the risk of bleeding.
With that caveat, I’m going to summarise the evidence for bleeds generally and then fatal ones before trying to hone in on bleeding just in the cancer post diagnosis circumstance.
Regarding general bleeds, low dose aspirin is associated with additional gastrointestinal (GI) bleeds in between 0.8 and 5.0 patients per 1,000 person years aged 50 to 84 years in the general population – the higher estimate being based on elderly subjects (Ref 5). This represents an increase above spontaneous GI bleeding in patients not taking aspirin of between about 50% (Ref 6) and 90% (Ref 7). Notwithstanding that the first of those results (50%) was related to aspirin and the prevention of atherosclerosis and the second (90%) was related to aspirin and the prevention of colon cancer, this suggests that only one in every two or every three bleeds that occur in patients taking low-dose aspirin is likely to be attributable to the aspirin, the other bleeds being spontaneous and not directly due to aspirin.
Regarding fatal bleeds, the go-to paper is another one by Elwood et al (Ref 8). It is a systemic review and meta-analysis of randomised trials that examined low dose aspirin and fatal bleeds. Eleven randomised trials were identified in the literature search – 10 were related to cardiovascular disease and only 1 to (colorectal) cancer and all were focused on prevention, not post-diagnosis treatment. The paper found that there was no significant increase in the risk of a fatal bleed among those randomised to aspirin (RR 0.77; 95% CI 0.41, 1.43). Four other reviews of low dose aspirin and fatal bleeds have also found no significant increase (Ref 9). One out of 28 studies in 1 of those 4 reviews was about (colorectal) cancer prevention. The rest were about cardiovascular disease and primarily about prevention, not treatment. Two other studies found lower relative risks attributable to aspirin, which is plausible if one surmises that those on aspirin might be more carefully monitored to catch potential bleeding issues early (Ref 10). One of these two studies was about aspirin for prevention of CVD events in the elderly. The other was relevant to our cancer treatment question. Rothwell et al found that “Case-fatality from major extracranial bleeds was lower on aspirin than on control (8/203 vs 15/132; OR 0·32, 95% CI 0·12-0·83, p=0·009).”
I did two further searches:
1) A pubmed search of meta-analyses (best evidence) published in the last five years (most complete and recent evidence) on the subject of low dose aspirin and bleeding. I did not specify fatal or otherwise.
2) A review of the number needed to treat website, which has independently amalgamated evidence to estimate the number needed to treat for an intervention to be effective and the number needed to harm from that intervention.
1) returned 38 papers (as of 1 September 2021) (Ref 11). Only two were relevant to our question. The others were on disparate topics e.g., one asked if one drug was better than another as a treatment if bleeding occurred, others compared aspirin with other anti-coagulants, others explored aspirin combined with other interventions, others were very specific to conditions or scenarios e.g., aspirin and orthopaedic operations, aspirin and in vitro fertilisation.
One of the two was very recent – it would have been published while the Elwood et al paper is in peer review (Ref 12). It was a meta-analysis of population studies (not as high evidence as meta-analysis of randomised trials). It was, however, pertinent to our narrow research question of cancer post-diagnosis. This found that decreased cancer-specific mortality was associated with post-diagnosis low-dose aspirin use (OR: 0.84; 0.75, 0.93). No data were presented in the paper on bleeding, so we may assume that it was not a significant issue.
The other wasn’t a meta-analysis per se. It was a review of systematic reviews and meta-analyses. It only focused on the use of low-dose aspirin as a prevention of CVD and the incidence of bleeding in this context. It presented its findings as number needed to treat and number needed to harm. It concluded that 138 people needed to be treated for 10 years to avoid 1 nonfatal MI, nonfatal stroke or CVD death and 108 needed to be treated over 10 years for 1 GI bleed (Ref 13). This doesn’t help with our cancer focus.
2) The number needed to treat web site has a section for oncology, but nothing in this section examined aspirin (Ref 14). I searched the NNT site generally for aspirin and eight references to aspirin were returned. All were in the subject area of CVD. The two that seemed most relevant were about aspirin to prevent a first heart attack or stroke, as it might be reasonable to assume that someone taking aspirin for cancer had not had a CVD event (given that most people have not had a CVD event). One review was done in January 2015 (Ref 15) and should be deleted, as it has been superseded by a more recent review (June 2019) (Ref 16). The June 2019 review reported that, taking aspirin over 5-7 years for CVD prevention, resulted in 1 in 250 people experiencing a major bleeding event. All evidence came from examinations of CVD, not cancer. This search elicited two more meta-analyses that were not cited in the Elwood et al paper awaiting publication and which may be of interest (Ref 17).
There are four major dimensions to aspirin research – cancer prevention and ‘treatment’ post-diagnosis and CVD prevention and ‘treatment’ post-event. Peter has focused on three of these – he has not examined aspirin for cancer prevention for two main reasons: i) another Peter (the Rothwell cited above) has and ii) Rothwell found that large and long studies would be needed to show benefit in this field. The paper being reviewed for this Monday note is for cancer post-diagnosis only. There is evidence of benefit. The usual 20% kind of relative benefit is never that impressive, but the days, weeks and months reported in the median survival time analysis were powerful. The plausible mechanisms are plentiful. The downside to consider is the risk of bleeding.
Low dose aspirin is associated with additional gastrointestinal (GI) bleeds in between 0.8 and 5.0 patients per 1,000 person years (with the elderly caveat for the upper limit). Comparison with normal bleeding suggests that only one in every two or every three bleeds that occur in patients taking low-dose aspirin is likely to be attributable to the aspirin. Fatal bleeds are the important factor. There’s no point delaying cancer death to die prematurely of a bleed. Elwood et al’s definitive paper on this found no significant difference in the risk of taking aspirin. Many other researchers have found the same. These papers were all exclusively or majorly about CVD and not cancer and about prevention and not post-diagnosis.
Elwood et al summarised their findings in the review of the 188 studies by saying “There is a considerable body of evidence suggestive of about a 20% reduction in mortality in patients with cancer who take aspirin, and the benefit appears not to be restricted to one or a few cancers. Aspirin, therefore, appears to deserve serious consideration as an adjuvant treatment of cancer, and patients with cancer, and their carers, have a right to be informed of the available evidence.”
I agree. Patients with cancer and their carers have a right to be informed of the available evidence. It is for them to decide – with the fundamental NHS principle of informed consent – whether or not to include low dose aspirin as part of their cancer treatment (Ref 18).
Ref 1: Fehily AM, Yarnell JWG, Sweetnam PM, et al. Diet and incident ischaemic heart disease: the Caerphilly Study. British Journal of Nutrition. 1993.
Ref 2: https://www.phpc.cam.ac.uk/event/does-aspirin-prevent-cancer-and-vascular-disease-40-years-of-data-and-controversy-professor-peter-elwood-cardiff-university/
Ref 3: Elwood et al. Aspirin and cancer survival: a systematic review and meta-analyses of 118 observational studies of aspirin and 18 cancers. ecancer 15 1258. 2021. https://ecancer.org/en/journal/article/1258-aspirin-and-cancer-survival-a-systematic-review-and-meta-analyses-of-118-observational-studies-of-aspirin-and-18-cancers
Ref 4: https://www.mdanderson.org/publications/focused-on-health/low-dose-aspirin-cancer-prevention.h20-1589046.html
Ref 5: Thorat et al. Prophylactic use of aspirin: systematic review of harms and approaches to mitigation in the general population. Eur J Epidemiol. 2013
Ref 6: Patrono et al. Low-dose aspirin for the prevention of atherosclerosis. N Engl J Med. 2008
Ref 7: Chubak et al. Aspirin Use for the Prevention of Colorectal Cancer: An Updated Systematic Evidence Review for the U.S. Preventive Services Task Force [Internet]. Rockville (MD): Agency for Healthcare Research and Quality (US); 2015 Sep. (Evidence Syntheses, No. 133.) https://www.ncbi.nlm.nih.gov/books/NBK321661/
Ref 8: Elwood et al. Systematic review and meta-analysis of randomised trials to ascertain fatal gastrointestinal bleeding events attributable to preventive low-dose aspirin: No evidence of increased risk. PLOS 2016 https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0166166
Ref 9: Antithrombotic Trialists’ Collaboration. Aspirin in the primary and secondary prevention of vascular disease: collaborative meta-analysis of individual participant data from randomised trials. Lancet. 2019. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2715005/
Lanas et al. Low doses of acetylsalicylic acid increase risk of gastrointestinal bleeding in a meta-analysis. Clinical Gastroenterol Hepatol. 2011. https://pubmed.ncbi.nlm.nih.gov/21699808/
McQuaid et al. Systematic review and meta-analysis of adverse events of low-dose aspirin and clopidogrel in randomised controlled trials. Amer J Med. 2006. https://pubmed.ncbi.nlm.nih.gov/16887404/
Wu et al. Case-fatality of recurrent venous thromboenbolism and major bleeding associated with aspirin, warfarin, and direct oral anticoagulants for secondary prevention. Thromb Res. 2015.
Ref 10: Rothwell et al. Short-term effects of daily aspirin on cancer incidence, mortality and non-vascular death: analysis of the time course of risks and benefits in 51 randomised trials. Lancet. 2012. https://pubmed.ncbi.nlm.nih.gov/22440946/
Mahady et al. Major GI bleeding in older persons using aspirin: incidence and risk factors in the ASPREE randomised controlled trial. Gut. 2021.
Ref 11: https://pubmed.ncbi.nlm.nih.gov/?term=low+dose+aspirin+bleeding&filter=pubt.meta-analysis&filter=datesearch.y_5
Ref 12: Wang et al. Low-dose aspirin use and cancer-specific mortality: a meta-analysis of cohort studies. J Public Health (Oxf). 2021 https://pubmed.ncbi.nlm.nih.gov/31781767/
Ref 13: Mutter et al. Clinical Inquiries: What are the benefits and risks of daily low-dose aspirin for primary prevention of CV events? J Fam Pract. 2018. https://pubmed.ncbi.nlm.nih.gov/29614147/
Ref 14: https://www.thennt.com/home-nnt/
Ref 15: https://www.thennt.com/nnt/aspirin-to-prevent-a-first-heart-attack-or-stroke-2/
Ref 16: https://www.thennt.com/nnt/aspirin-preventing-first-heart-attack-stroke/
Ref 17: Mahmoud et al. Efficacy and safety of aspirin for primary prevention of cardiovascular events: a meta-analysis and trial sequential analysis of randomized controlled trials. Eur Heart J. 2019. and
Zheng et al. Association of Aspirin Use for Primary Prevention With Cardiovascular Events and Bleeding Events: A Systematic Review and Meta-analysis. JAMA. 2019. https://pubmed.ncbi.nlm.nih.gov/30667501/
Ref 18: https://www.nhs.uk/conditions/consent-to-treatment/