To determine whether aspirin use affected risk for hepatocellular carcinoma (HCC), a recent meta-analysis examined 18 cohort and case-control studies published through September 2021. Included studies assessed the association between aspirin versus another medication, no medication, or placebo and liver cancer in adults. The study found that patients who used aspirin were less likely to develop hepatocellular carcinoma than those who did not and that aspirin was associated with a protective effect against HCC after hepatitis B virus infection and hepatitis C virus infection, as well as in patients with chronic liver disease and in the general population. The authors concluded that aspirin was associated with lower HCC risk.
The study was published online March 22 by the Journal of Clinical Gastroenterology. The following commentary by Ronald L. Koretz, MD, FACP, was published in the ACP Journal Club section of the July Annals of Internal Medicine.
The systematic review and meta-analysis by Wang and colleagues found that aspirin use was associated with a reduced risk for HCC. However, there are some problems with the analyses.
Most of the 18 studies (13 by my estimate) used the same databases, substantially reducing the independence of the estimates. For example, a Taiwan database and a Korean database were each used in 4 studies that were included in the review. Some of the odds ratios used in the primary meta-analysis differed from the numbers in the original papers. Most of these differences were trivial, but the reported hazard ratio in Lee and colleagues' article (1) was 0.78, not 0.31.
Cohort studies can be based on planned, new data collection (prospective) or constructed from data already collected for a different purpose (retrospective). Prospective data collection is much preferred because the investigators can assure its relevance and quality for their study questions. It is not clear what the 7 studies in the review called only “cohort” represented. Most of these cohort studies appeared to be constructed retrospectively. Wang and colleagues focused on the heterogeneity between studies but, because all of the estimates in the studies were in the same direction, this may not have been a major issue. However, the systematic review did not retrieve 1 case-control study that should have met inclusion criteria, and it did not show an association between aspirin and HCC.
Despite the problems, the results of the meta-analyses do provide support for an association between aspirin use and reduced risk for HCC. However, association does not mean causation. Observational studies cannot establish causation because of confounding. Only 2 of the 9 Bradford Hill criteria for establishing causation from observational studies are clearly present here: consistency among different studies and temporality (exposure precedes event). The authors argue that the association is biologically plausible, but the evidence suggesting that aspirin prevents other malignancies requires support from randomized controlled trials.
The current evidence seems to support an association between aspirin and lower risk for HCC but falls short of warranting use of aspirin as prophylaxis for HCC. Further exploration is needed.