Nonsteroidal anti-inflammatory drug use, chronic liver disease, and hepatocellular carcinoma: The egg of columbus or another illusion?


  • Amy K. Kim M.D.,

    1. Yale University Liver Center, Department of Internal Medicine, Yale University School of Medicine, New Haven, CT
    Search for more papers by this author
  • James Dziura PH.D.,

    1. Department of Emergency Medicine, Yale Center for Analytical Sciences, Yale University School of Medicine, New Haven, CT
    Search for more papers by this author
  • Mario Strazzabosco M.D., PH.D.

    1. Yale University Liver Center, Department of Internal Medicine, Yale University School of Medicine, New Haven, CT
    2. Digestive Disease Section, Department of Surgical and Medical Translational Sciences, University of Milan-Bicocca, Milan, Italy
    Search for more papers by this author

  • Potential conflict of interest: Nothing to report.

Sahasrabuddhe VV, Gunja MZ, Graubard BI, Trabert B, Schwartz LM, Park Y, et al. Nonsteroidal anti-inflammatory drug use, chronic liver disease, and hepatocellular carcinoma. J Natl Cancer Inst 2012;104:1808-1814. (Reprinted with permission.)


Background: Nonsteroidal anti-inflammatory drugs (NSAIDs) have been shown to reduce chronic inflammation and risk of many cancers, but their effect on risk of hepatocellular carcinoma (HCC) and death due to chronic liver disease (CLD) has not been investigated. Methods: We analyzed prospective data on 300 504 men and women aged 50 to 71 years in the National Institutes of Health–AARP Diet and Health Study cohort and linked self-reported aspirin and non-aspirin NSAID use with registry confirmed diagnoses of HCC and death due to CLD. We calculated hazard rate ratios (RRs) and their two-sided 95% confidence intervals (CIs) using Cox proportional hazard regression models with adjustment for age, sex, race/ethnicity, cigarette smoking, alcohol consumption, diabetes, and body mass index. All tests of statistical significance were two-sided. Results: Aspirin users had statistically significant reduced risks of incidence of HCC (RR = 0.59; 95% CI = 0.45 to 0.77) and mortality due to CLD (RR = 0.55; 95% CI = 0.45 to 0.67) compared to those who did not use aspirin. In contrast, users of non-aspirin NSAIDs had a reduced risk of mortality due to CLD (RR = 0.74; 95% CI = 0.61 to 0.90) but did not have lower risk of incidence of HCC (RR = 1.08; 95% CI = 0.84 to 1.39) compared to those who did not use non-aspirin NSAIDs. The risk estimates did not vary in statistical significance by frequency (monthly, weekly, daily) of aspirin use, but the reduced risk of mortality due to CLD was statistically significant only among monthly users of non-aspirin NSAIDs compared to non-users. Conclusions: Aspirin use was associated with reduced risk of developing HCC and of death due to CLD whereas nonaspirin NSAID use was only associated with reduced risk of death due to CLD.


Hepatocellular carcinoma (HCC) imposes an enormous burden in terms of morbidity and mortality and their associated costs. The incidence and prevalence of HCC are increasing also in Western countries, where HCC is now the leading cause of death in patients with liver cirrhosis. Implementation of surveillance protocols have improved the prognosis of the treated patients but, unfortunately, more than 80% of HCC is diagnosed in areas lacking adequate infrastructures, leaving the vast majority of the patients without proper treatment. In the U.S., more than 50% of patients still remain untreated.

Several treatment options are available for patients with early to intermediate disease. These are often used sequentially and the costs of HCC management are elevated, compared to other neoplasm. Because of the aggressiveness of the disease, the unsatisfactory access to proper care and the costs associated with HCC management, major efforts should be made in the implementation of preventative measures.

Vaccination for hepatitis B virus (HBV) is available and it has been shown to decrease the incidence of HCC in populations with endemic HBV infection. Measures to effectively prevent HBV/HCV infection as well as alcoholic liver disease and metabolic liver disease are well known; however they require modification of life style and are slow to become effective. In addition, alcohol consumption in the younger generations and in countries that previously had a more moderate intake is actually on the rise, clearly becoming a matter of great concern.

Eradication of HCV and the life-long use of antivirals with high biological barrier reduce the incidence of HCC in HCV- and HBV-infected patients, respectively. When etiologic treatment in patients with chronic liver disease (CLD) is not available or fails, prevention of HCC aims at halting necroinflammation and fibrosis.

In this scenario, chemoprevention strategies with drugs that are able to target common pathogenic mechanisms are of great interest. One such strategy is the use of aspirin. The role of aspirin in HCC and CLD was addressed in two recent studies. The first article, published in 2012 by Sitia et al.,[1] showed that administration of antiplatelet treatment (aspirin or clopidogrel) to rodents with chronic necroinflammatory liver disease related to HBV reduced the severity of liver fibrosis and the development of HCC, along with a reduction of HBV-specific CD8+ T cells and of HBV-nonspecific inflammatory cells.

The second article, from Sahasrabuddhe et al.,[2] is an epidemiological study based on a very large cohort from the National Institutes of Health (NIH) American Association of Retired Persons (AARP) Diet and Health Study database, analyzing the survival in self-reported use of aspirin or nonaspirin nonsteroidal antiinflammatory drug (NSAID) in patients with HCC and chronic liver disease. The study enrolled 300,504 patients who completed a survey at baseline from 1995-1996, and then a follow-up “risk-factor” survey 6 months later, which included information on the use of NSAIDs. Primary HCC incidence was identified through linkage with state cancer registries and the National Death Index, until 2006. Patients who died from chronic liver disease were also identified until 2008. Hazard ratios were calculated using a Cox regression model after adjusting for age, sex, race, cigarette smoking, alcohol consumption, diabetes, and body mass index (BMI).

The authors found that of the 250 patients with HCC and 428 deaths from CLD, aspirin use was associated with reduced risk of developing HCC and death due to CLD, whereas use of nonaspirin NSAID was associated with only reduced death from CLD. The use of aspirin alone was associated with a 41% reduced risk of developing HCC and a 51% reduced risk of death from CLD not related to HCC. The users of only nonaspirin NSAID were also at reduced risk of death for CLD by 34%, but the risk of developing HCC remained unchanged.

There is biological plausibility to these associations; in fact, the role of the inflammatory field in liver cirrhosis and cancer has been suggested by a number of studies, stemming from the knowledge that liver cancer occurs in the setting of a chronic inflammatory state.

The effects of aspirin and nonaspirin NSAID rely on several possible mechanisms, only partly related to the inhibition of Cox enzymes and prostaglandins, such as a decrease in angiogenesis and in cancer cell proliferation, increase in apoptosis, and reduction in proinflammatory cytokine production.

Sitia et al.[1] found that the effects of aspirin were present in HBV-related carcinogenesis, but not in chemical liver carcinogenesis. Most of the cases of HCC in patients affected with HBV occur after years of immune-mediated necroinflammation characterized by functionally ineffective virus-specific CD8+ T-cell response that fails to eliminate the virus from the liver. In this situation, the virus-specific CD8+ T-cell response maintains a cycle of low-level liver injury/inflammation through the recruiting of virus-nonspecific inflammatory cells that exacerbate the inflammatory reaction. Aspirin would be able to interfere with the recruitment of this second component of inflammatory cells, thereby reducing inflammation. Furthermore, platelet inhibition may block the release of important growth factors, such as FGF, HGF, ILF, VEGF, PDGF, and serotonin, that play a role in HCC development and growth. Unfortunately, Sahasrabuddhe et al.'s epidemiological study did not provide data on the stratification of the protective effect according to the causes of CLD and HCC, hence lacking confirmation that the aspirin effect is selective for HBV-related liver disease.

However, an antiviral activity of aspirin against HCV or other flaviruses has already been suggested, by way of COX-2 inhibition,[3] and by way of the induction of Cu/Zn-SOD expression as well as direct antioxidant properties.[4]

COX-independent, platelet independent and antioxidant-independent protective effects of aspirin against liver injury have also been reported. Imaeda et al.,[5] using an acetaminophen-induced acute liver injury model, showed that low-dose aspirin inhibits inflammasome-mediated pathways, thereby reducing the transcription of inflammatory cytokines.

Chemoprevention of cancer with aspirin is not a novel concept. It has been investigated in the setting of colorectal cancer through multiple cohort and case control studies, demonstrating benefit; however two large randomized controlled trials (RCTs)[6, 7] that included more than 22,000 and 39,000 patients, respectively, did not show significant benefit in reducing colorectal cancer incidence. Thus, a final judgment on the effect of aspirin on colorectal cancer prevention is still pending.[8]

Similarly, the evidence presented in Sahasrabuddhe et al.'s[2] study is not robust enough to recommend the use of aspirin in the prevention of HCC. The strengths of the study included the statistical power from the large cohort with many events, the ability to separate aspirin from nonaspirin NSAIDs, and the robustness of results to sensitivity analyses addressing protopathic bias and confounding by indication. Despite the striking results from this study as well as other studies providing support for biological plausibility, caution must be taken in their interpretation. There are several well-documented examples of the inability to reproduce associations from observational studies into clinical trial settings, including the Women's Health Study[7] which was unable to detect a benefit of low-dose aspirin on cancer. In other words, observing that aspirin users are less likely to develop HCC than nonusers does not necessarily mean that giving aspirin to patients will reduce their likelihood of HCC. Several factors may underlie this discrepancy, including variability in study populations, dose and duration of intervention, and differential measurement. However, a fundamental challenge for observational studies is the opportunity for selection bias. Healthy-user bias exists when people who faithfully engage in activities that are good for them (e.g., using aspirin daily to reduce cardiovascular disease [CVD]) are fundamentally different from those who do not. In fact, self-selected aspirin use has been shown to be associated with factors predictive of cancer.[6-8] While documented risk factors for HCC were similar between the aspirin users and nonusers in the AARP cohort, several other factors known to be associated with cancer and mortality were not assessed, including socioeconomic status, diet, and physical activity.

A biologic gradient is one of the nine criteria for causation proposed by Sir Bradford Hill. In the AARP study, those reporting monthly aspirin use received the same benefit as those reporting daily or weekly use. Certainly the imperfect self-reported measurement of frequency of aspirin use along with the concentration on only the previous year's aspirin exposure could have hindered the detection of a dose-response relation. Nevertheless, the potential benefit that amounts from only occasional (i.e., monthly) use is inconsistent with biological plausibility arguments that are directed at the reduction of chronic inflammation.

Aspirin, statins, and metformin are well-known medications used to treat metabolic and cardiovascular disease and seem to find indications in preventing and treating chronic liver disease and liver cancer. Confounding by indication is another possible explanation for the observed association. It is possible that aspirin is given to a population of patients in which cirrhosis and HCC are less progressive. Perhaps aspirin is more likely to be prescribed to patients with metabolic syndrome rather than HCV-related liver disease.

Ancient cultures placed the site of life and sentiment in the liver, considering it to be the central organ of the human body, the seat of life, soul, and intelligence[9]; more recently, poets and story tellers identify the heart as the seat of sentiment and emotions. What is good for the heart may turn out to be also good for the liver, but for aspirin, better evidence is needed.

  • Amy K. Kim, M.D.1James Dziura, Ph.D.2Mario Strazzabosco, M.D., Ph.D.1,3

  • 1Yale University Liver CenterDepartment of Internal MedicineYale University School of MedicineNew Haven, CT2Department of Emergency MedicineYale Center for Analytical SciencesYale University School of MedicineNew Haven, CT3Digestive Disease SectionDepartment of Surgical and Medical Translational SciencesUniversity of Milan-BicoccaMilan, Italy