crabsallover highlights, key points, comments / links.
The Lancet, Volume 377, Issue 9778, Pages 1649 - 1650, 14 May 2011
doi:10.1016/S0140-6736(11)60666-1
Cite or Link Using DOI
Cite or Link Using DOIAspirin in the prevention of cancer
We write with regard to the Article by Peter Rothwell and colleagues,1 which indicates that low-dose aspirin might prevent cancer. Although we strongly endorse Rothwell and colleagues' publication, we urge caution.
If we presume that aspirin does have a chemoprotective role, it clearly does not work for everyone since at best 25% of people are estimated to get a cancer prevention benefit but this figure could be as low as 20%. This “aspirin resistance” in patients seems to be present widely in the population so we do not know who should respond.2 Genetic studies such as those undertaken by the Esophageal Adenocarcinoma Genetic Linkage (EAGLE) consortia are needed to assess which individuals will respond best to the chemopreventive effects of low-dose aspirin.
Additionally, meta-analysis of cardiac trials with reassessment of causes of cancer deaths might have inadvertently introduced bias. For example, the number of cases of gastrointestinal cancer death in Rothwell and colleagues' study was 182 out of almost 20 000 cases. In particular there were only 23 oesophageal cancers. Patients in these trials, especially those using aspirin, might have had complications that resulted in earlier presentation.
The AspECT chemoprevention trial3 was specifically designed to look at the effect of aspirin on oesophageal cancer development but will also give information on effects on colon cancer development as well as on cardiac deaths. To date, the data monitoring team and the trial steering team have not divulged any obvious trend between the four groups of this trial (low-dose proton-pump inhibitor [PPI], high-dose PPI, low-dose PPI with low-dose aspirin [300 mg], and high-dose PPI with low-dose aspirin). We need trials such as AspECT to report their preliminary findings of genetic stratification for response, as well as risk/benefit, dose, and length of therapy.
JJ has a consultancy to AstraZeneca Oncology, which makes esomeprazole—one of the agents in the AspECT trial. JJ has also received grants in aid from AstraZeneca previously. All authors are co-investigators in the AspECT trial.
References
1 . Effect of daily aspirin on long-term risk of death due to cancer: analysis of individual patient data from randomised trials. Lancet 2011; 377: 31-41. Summary | Full Text | PDF(502KB) |CrossRef | PubMed
a Leicester Royal Infirmary, Leicester LE7 7HH, UK
b Gloucestershire Royal Hospital, Gloucester, UK
c Belfast Hospital Trust, Belfast, UK
d Northumbria Healthcare NHS Foundation Trust, Tyne and Wear, UK
e Gastrointestinal Division, McMaster University Medical Centre, Hamilton, ON, Canada
We thank Janusz Jankowski and colleagues for their comments. However, our estimate of the effect of aspirin on death due to oesophogeal cancer was based on 68 events and not 23 as Jankowski and colleagues state. The 20-year risk of death due to adenocarcinoma of the oesophagus, the predominant histological subtype in Barretts' oesophagus, was substantially reduced by allocation to aspirin in the trials (hazard ratio 0·36, 95% CI 0·21—0·63, p=0·0001).
It is difficult to conceive of how this estimate might have been biased by inclusion of some trials that included patients with established vascular disease, as Jankowski and colleagues suggest. Moreover, the effect was significant in both of the large randomised trials of daily aspirin versus control in primary prevention that we studied—ie, the Thrombosis Prevention Trial1(0·42, 0·22—0·78, p=0·004) and the British Doctors Aspirin Trial2 (0·45, 0·19—0·99, p=0·04). The figure shows a pooled analysis of the effect in these two trials.
Additionally, our estimates of the effect of aspirin on risk of death due to gastrointestinal cancers as a whole were based on more than 500 deaths and not on 182 as stated by Jankowski and colleagues, with a highly statistically robust reduction in 20-year mortality (0·65, 0·53—0·78, p<0·0001).
No comments:
Post a Comment