Skip navigation
Link to Back issues listing | Back Issue Listing with content Index | Subject Index

Low-dose aspirin and cancer

Study [1]
Results
Comment

Past observational studies have noted a tendency for lower cancer rates in people who took aspirin or NSAIDs on a regular basis. On the basis of this, trials have been performed to see if aspirin, and, later, coxibs, prevented or delayed colorectal polyp development. The choice of colorectal polyps was because this was a surrogate marker of cancer prevention, in studies that could be done relatively quickly, over one to three years. The results were summarised in Bandolier 129, showing small effects if any.

Real clarity would only come from large, long-term studies, especially in primary prevention. Knowing the answer is important. In the USA in particular, much has been written in the media in recent months suggesting that an aspirin a day cures all known ills.

Not so, as we know that aspirin at any dose is associated with real harm (Bandolier 122). We now have clarity. The bottom line is that in people without previous cancer, low dose aspirin has no effect over cancer development.

Study [1]

A large number of female healthcare professionals who were at least 45 years old and without previous history of cancer, cardiovascular disease, or other major chronic illness formed the initial cohort. These were randomised to 100 mg aspirin every second day, or placebo. They had not to be taking aspirin or NSAIDs regularly. Enrolment was between mid 1992 to mid 1995, and the end of the trial was in 2004.

Every six months initially, and then annually, participants were sent questionnaires asking about compliance, adverse events, occurrence of endpoints, and risk factors. If an endpoint was reported, medical records were sought and reviewed by a panel of assessors blinded to treatment assignment. Only confirmed cancer endpoints were used.

Results

Just fewer than 40,000 women were randomised. Follow up on morbidity and mortality was close to 100%, and compliance averaged 73%. Women had an average age of 55 years, a mean BMI of 26, and about 13% smoked. About 18% had a family history of cancer. Treatment groups were identical at baseline.

Taking aspirin had no effect on cancer rates, in total or by individual cancer, or cancer type, or on cancer deaths, or on all-cause mortality. No difference was found for any subgroup, for instance those with a family history of cancer, or who smoked.

Over the average 10.1 years of follow up, 2,865 women developed cancer, a rate of just over 7%, or 1 in 14 women (Table 1). About 3% of women died over the 10 years, about 1 in 32, and under half, about 1 in 68, were cancer deaths.



Table 1: Ten-year risk of cancer in female healthcare professionals of average age 55 years (all women in trial)



 

Number of women with the cancer

 

Percent of all women

10-year risk

(1 woman in )

Total cancers
2865
7.18
14
Total cancer deaths
583
1.46
68
Total deaths
1251
3.14
32
Individual cancer by site
Breast
1230
3.08
32
Colorectal
269
0.67
147
Uterine
224
0.56
176
Lung
205
0.51
193
Other sites
151
0.38
261
Melanoma
138
0.35
286
Non-Hodgkin lymphoma
133
0.33
297
Ovary
129
0.32
306
Leukaemia
61
0.15
647
Kidney
61
0.15
647
Thyroid
60
0.15
658
Pancreas
51
0.13
774
Baldder
51
0.13
774
Brain
31
0.08
1274
Multiple myeloma
30
0.08
1316
Stomach
20
0.05
1974
Hodgkin lymphoma
11
0.03
3589
Oesophagus
7
0.02
5640


The most common cancers were breast cancer (in 3%), colorectal (0.7%), uterine (0.6%) and lung (0.5%). All other cancers occurred in under half of one percent of women. The overall individual risk of developing a particular cancer over 10 years ranged from 1 in 32 for breast cancer to 1 in 5600 for oesophageal cancer.

Comment

This is an important trial because it tests an important hypothesis in a large number of women over a long period of time, and with a sufficient number of events to be sure of the result. In people with no history of cancer, low-dose aspirin does not prevent cancer. Low dose aspirin is not without harm, however, so on balance in these patients it probably does more harm than good.

Another reason it is important is because it demonstrates that however many observational studies may have suggested there was a connection between aspirin or NSAID consumption and lower rates of cancer, they were almost certainly wrong. Whatever was going on, it wasn't the aspirin or NSAID causing lower cancer rates.

It may be that higher aspirin doses might have some effect, and here the dose was only 50 mg a day on average. It may be that inhibition of different cyclooxygenase enzymes (cox-2 rather than cox-1) may have an effect, but there is controversy about this because cox-2 inhibition and aspirin caused heart attacks or strokes in polyp prevention studies.

The lesson for all of us is that however good an observational study may be, it can still be wrong, especially in complex situations where there is limited effect and possible unknown confounding factors.

Reference:

  1. NR Cook et al. Low-dose aspirin in the primary prevention of cancer. JAMA 2005 294: 47-55.

previous or next story