Proton pump inhibitor use associated with an increased risk of gastric cancer

Référence

Abrahami D, McDonald EG, Schnitzer ME, Barkun AN, Suissa S, Azoulay L. Proton pump inhibitors and risk of gastric cancer: population-based cohort study. Gut 2022;71(1):16-24.

Sommaire

Proton pump inhibitors (PPIs) cause hypergastrinemia, which can lead to hyperplasia of the gastric mucosa, and several previous studies have shown an association between PPI use and gastric cancer. This study is the largest and most methodologically sound; it addresses several potential sources of confounding better than previous studies. The authors identified more than 1.1 million persons who had received a new prescription for a PPI between 1990 and 2018, and another 220,825 persons who had received a new prescription for a histamine-2 receptor antagonist (H2RA) during the same period. This is a better comparison group than the general population, since it avoids confounding by indication (PPI users may have different risk factors and health habits than non-users). Persons with a familial syndrome associated with gastric cancer, persons with previous gastric cancer, and those who had less than one year of follow-up were excluded, leaving 973,281 PPI patients and 198,306 H2RA patients. The authors matched patients using propensity scores that incorporated a large array of potential confounders, including age, comorbidities, tobacco and alcohol use, and medications. In the fully adjusted model, the authors found a significantly increased risk of gastric cancer in PPI users compared with H2RA users (hazard ratio [HR] = 1.45; 95% CI 1.06 - 1.98; number needed to treat to harm = 2121 after 5 years and 1191 after 10 years). The Kaplan-Meier survival analysis showed that the risk increased linearly with the duration of PPI use. Greater doses were associated with increased risk: HR = 1.22 for less than 14,600 mg cumulative omeprazole dose equivalents; HR = 1.81 for 14,600 mg to 28,199 mg dose equivalents; and HR = 2.03 for 29,200 mg and higher dose equivalents (although the confidence intervals overlap).