TL;DR
This systematic review of randomized controlled trials found that while most NSAIDs have very low rates of significant liver enzyme elevations, diclofenac and rofecoxib showed higher rates compared to placebo. Importantly, despite these elevated liver enzymes, no NSAID demonstrated an increased risk of serious liver-related adverse events, hospitalizations, or deaths.
What Problem Does This Paper Address?
Nonsteroidal anti-inflammatory drugs (NSAIDs) are widely used for pain and inflammation, but concerns about their potential for liver toxicity persist. While rare, serious hepatic injury, including acute liver failure, has been associated with some NSAIDs, leading to their withdrawal from the market. However, the precise frequency of these laboratory and clinical side effects across commonly used NSAIDs, especially in the context of arthritis patients, remained unclear. Individual studies are often too small to reliably estimate these risks. This paper aims to systematically review existing randomized controlled trials to provide a clearer picture of NSAID-induced hepatic toxicity.
Methods
The researchers conducted a comprehensive systematic review of randomized controlled trials (RCTs) involving adults with osteoarthritis or rheumatoid arthritis. They searched major bibliographic databases (MEDLINE, EMBASE) and public archives of the Food and Drug Administration (FDA) for studies reporting on specific NSAIDs (diclofenac, naproxen, ibuprofen, celecoxib, rofecoxib, valdecoxib, or meloxicam).
Key inclusion criteria for studies were:
- RCTs of adults (age > 18 years).
- Study duration of 4 weeks or greater.
- At least one of the specified NSAIDs used.
- Study arms with more than 40 patients.
- Explicit reporting of specific hepatic toxicity outcomes:
- Aminotransferase elevations >3x the upper limit of normal (ULN).
- Liver-related study discontinuations.
- Liver-related serious adverse events (SAEs).
- Liver-related hospitalizations.
- Liver-related deaths.
Data abstraction and quality assessment were performed independently by two investigators to ensure rigor. Statistical analysis involved sample size-weighted pooling of data for each NSAID and outcome to calculate proportions and their 95% confidence intervals (CIs). Subgroup analyses were conducted based on study duration and NSAID dose.
Datasets
The study did not introduce new datasets. Instead, it synthesized data from 67 published articles and 65 studies from FDA archives that met the inclusion criteria. These existing studies, primarily RCTs in arthritis patients, formed the basis of the review.
Key Insights and Ideas
- Differential Hepatic Risk: Not all NSAIDs are created equal when it comes to liver enzyme elevations. While most showed rates comparable to placebo, diclofenac and rofecoxib stood out with significantly higher rates of aminotransferase elevations.
- Enzyme Elevation vs. Serious Harm: The most striking finding is the disconnect between laboratory indicators of liver stress (elevated enzymes) and actual clinical harm. Even with higher enzyme elevations, the risk of serious liver-related events, hospitalizations, or deaths remained extremely low across all NSAIDs.
- Diclofenac's Dual Nature: Diclofenac not only showed higher aminotransferase elevations but also a higher rate of study discontinuations due to liver issues, suggesting a more pronounced impact on liver function markers and patient management.
- Dose and Duration Matter (for Diclofenac): For diclofenac, higher doses and longer durations of use were associated with an increased risk of aminotransferase elevations, highlighting a dose-response relationship for this specific NSAID.
- Limitations of "Hy's Rule": The study noted the difficulty in systematically assessing "Hy's rule" (concomitant elevations in aminotransferases and bilirubin) due to a lack of reporting in the included studies, underscoring a gap in how hepatic toxicity is documented.
Key Results
- Aminotransferase Elevations (>3x ULN):
- Placebo: 0.29% (95% CI, 0.17%-0.51%)
- Diclofenac: 3.55% (95% CI, 3.12%-4.03%)
- Rofecoxib: 1.80% (95% CI, 1.52%-2.13%)
- Other NSAIDs (ibuprofen, naproxen, meloxicam, celecoxib, valdecoxib): Rates ≤0.43%, with CIs overlapping placebo.
- Liver-Related Study Discontinuations:
- Placebo: 0.08% (95% CI, 0.02%-0.29%)
- Diclofenac: 2.17% (95% CI, 1.78%-2.64%)
- Other NSAIDs: CIs overlapped with placebo.
- Liver-Related Serious Adverse Events (SAEs): Very rare across all NSAIDs, with rates no higher than 0.06% for naproxen, rofecoxib, and diclofenac, and CIs overlapping placebo.
- Liver-Related Hospitalization & Death: Only one hospitalization and one death were reported in total across tens of thousands of patients, both associated with naproxen. The rates were extremely low (2.7 per 100,000 for hospitalization, 1.9 per 100,000 for death), with 95% CIs overlapping placebo.
Limitations and Future Work
The authors acknowledge that their analysis, being a compilation of different studies rather than direct head-to-head comparisons, limits direct statistical comparisons between NSAIDs. They also note that while elevated aminotransferases are an early indicator, their study might have lacked sufficient power to definitively link these elevations to rare but serious clinical liver events.
Future work should focus on identifying reliable methods to predict which medications cause liver failure at the pre-approval stage. Early, systematic pharmacovigilance programs using large clinical databases are crucial for identifying drugs that cause serious hepatic injury, even if rarely. The authors also suggest that further research is needed to systematically assess the relationship between elevated aminotransferase levels and serious clinical liver events, potentially requiring larger patient cohorts. The lack of reporting on concomitant bilirubin elevations ("Hy's rule") in many studies also highlights a need for more standardized reporting of hepatic safety data.
Summary updated