Proton pump inhibitors and gastric cancer: a population-based cohort study

Abstract

We read with great interest the recent article by Seo et al regarding the association of proton pump inhibitors (PPI) and gastric cancer. We want to congratulate the authors for their exemplary work as the prevalence of PPI in current clinical practice cannot be overstated. After reviewing the article, we have some comments to share with you. As mentioned in the Methods section and also in figure 1, the target cohort was defined as ‘new PPI users’ who were prescribed PPI for more than 30 consecutive days. Although we understood that the subjects underwent 365 days of continuous observation to be classified as new PPI users, we are curious to know if the remote history of PPI use (>365 days ago) was taken into consideration before enrolment into the target cohort. The authors have defined index date as the first day of a new drug prescription but sometimes patients may not start to take these medications right away after the prescription, thus, it is difficult to match this variable in both groups. Similarly, we were interested to know if gaps in medication prescription refills were matched appropriately between target and comparative cohorts. There is a possibility that the study population in the comparative cohort may have a longer gap between their subsequent prescription refills than the target cohort, leading to inequality in the total duration of drug exposure in both groups. Furthermore, any new drug user other than PPI was granted entry to the comparative cohort after meeting eligibility criteria, but there is no further information in the article regarding these drugs that the subjects in the comparative cohort were taking. This information would allow for a better comparison between PPI and other drugs. As mentioned under exclusions in figure 2A, 4093 people were excluded for being simultaneously included in both cohorts. Although it is correct to exclude them, we are interested to know how somebody who is taking PPI got included in the comparative cohort at the same time, as they clearly should not meet the eligibility criteria for enrolment. In the secondary analysis for identifying the effect of PPI use on gastric cancer development in the Helicobacter Pylori eradicated population, the index date was defined as the first day of H. Pylori eradication. Although eradication status was identified through the prescription of a combination drugs in the charts, there is no information on whether they considered the first day of the prescription or the last day of the eradication treatment as the index date. While the authors have mentioned in the discussion section that they could not determine H. Pylori infection status by a laboratory test, it is important to know if both groups were matched regarding clarithromycin based triple therapy or bismuthbased quadruple therapy, as they might differ in efficacy considering 20% resistance rate to clarithromycin in Korea. The authors also mentioned a cumulative dose–response relationship between increasing PPI duration and the risk of gastric cancer development. Though it is true for PPI≥30 and≥180 days, the same is not true for PPI≥365 days, as the result was not statistically significant (PPI≥365 days, HR 3.5, 95% CI 0.85 to 23.49, p value 0.14). Summarily, the gastric cancer incidence rate is higher in Korean men than women when we compare ≥40 years of age groups. Hence, it is important to know if gender was equally distributed for each age range in table 1. Our concern is if the PPI group had more men in ≥40 years age group, it can skew the result in the favour of PPI.