2018 Section 6 - Laryngology, Voice Disorders, and Bronchoesophalogy

Long-term safety of PPI therapy

LOTUS studies, some discussion of observed differences is perhaps warranted. Plausible explanations for the higher overall frequency of reported SAEs in the SOP- RAN study compared with the LOTUS study include the two-fold higher median exposure time in the SOPRAN study compared with the LOTUS study, and the fact that patients were older on average in the former study (med- ian: 51 – 55 vs. 45 years). 8, 10 The slightly higher levels of gastrin increase seen in the LOTUS study compared with the SOPRAN study is probably caused by the differences in trial design as the LOTUS protocol allowed for a planned symptom-driven dose escalation to 20 mg twice daily if needed, which is known to result in higher acid suppression than 40 mg twice daily. 12 A dose of 20 mg twice daily was required by 5 years in 13% of patients in the LOTUS study. There were few fatal SAEs in either group across both studies and the causes of death, which included cardiac disorders, pneumonia, and pancreatic or lung neoplasms, were considered by the treating physicians to be unre- lated to study treatment. It should be noted that surgery- speci fi c AEs (e.g. postoperative infections) will inevitably accumulate during the early part of the follow-up period and should be factored into long-term comparisons with drug safety. This important consideration has been care- fully incorporated into the present analyses. The overall occurrence of SAEs over time was similar for patients in the two studies irrespective of whether they were allocated to either PPI or ARS therapy. How- ever, myocardial infarction was reported more com- monly in the omeprazole than in the open ARS group in the SOPRAN study. This numerical imbalance was com- municated to regulatory authorities worldwide and is also presented in a previous paper by Lundell et al. (2009). 8 The data were evaluated by the FDA which con- cluded that this difference could have been driven by baseline differences [e.g. the higher median age (54 vs. 50 years) in the omeprazole group compared with the open ARS group, the higher frequency of previous myo- cardial infarction ( n = 6 vs. n = 0)] and other confound- ing factors [e.g. higher withdrawal from the open ARS group after randomisation but before treatment, longer median exposure time for the omeprazole group com- pared with the ARS group (12 vs. 10 years)], making it dif fi cult to draw conclusions regarding potential associa- tions between omeprazole treatment and the develop- ment or worsening of heart problems. In 2007, the FDA completed their comprehensive, scienti fi c review of com- plementary safety data for omeprazole as well as esomeprazole, including those provided by the SOPRAN

pectoris; patient had a history of angina pectoris before omeprazole therapy was initiated).

DISCUSSION The ideal control group for an assessment of long-term PPI therapy should have the same disease and baseline characteristics as the intervention group with the only difference being the absence of acid-suppressive therapy. In reality, such a control group cannot be ethically jus- ti fi ed, as it would mean withholding therapies that are known to be effective in patients with GERD. The SOPRAN and LOTUS studies provide the next best alternative; prospective, randomised data on the safety of long-term therapy with esomeprazole and omepra- zole in patients with chronic symptomatic GERD in comparison with a relevant control group (patients with the same disease states but allocated to ARS). While the frequency of certain SAEs was higher in the PPI group than in the ARS group in some cases (and vice versa), no trends were consistently observed across both studies, and may in most cases be due to natural varia- tion in background event rates (i.e. unrelated to treat- ment and thus not a true effect). Furthermore, certain limitations must be acknowledged when comparing treatment groups in open studies such as SOPRAN and LOTUS. While all patients were suitable candidates for ARS or PPI treatment of chronic GERD, in both stud- ies more patients randomised to ARS eventually refused treatment than those randomised to PPI treatment. 8, 10 Factors in fl uencing patients ’ decisions not to proceed with ARS could have also in fl uenced their risk of hav- ing certain SAEs (i.e. confounding). Differences in base- line patient characteristics and exposure times (discussed in detail below) may also have in fl uenced the frequency of certain SAEs between treatment groups. Statistical tests comparing the incidence of SAEs were not performed because the original studies were suf fi ciently powered only to detect differences in treatment failure and the sample sizes were not large enough to evaluate whether there were statistically sig- ni fi cant differences in SAEs between treatment groups. Furthermore, using a statistical test for each AE term would lead to a multiple testing problem, increasing the risk of fi nding false signi fi cant differences with the number of tests. Overall, our results do not raise any safety concerns with esomeprazole and omeprazole therapy when given daily at therapeutic doses to patients with GERD for 5 – 12 years. While it is not appropriate and was not our intention to compare safety data between the SOPRAN and

Aliment Pharmacol Ther 2015; 41: 1162 – 1174 ª 2015 John Wiley & Sons Ltd

184