Abstract
We thank Dr Khouri et al. for their interest in our article titled “Exploring the association between weight loss-inducing medications and multiple sclerosis: insights from the FDA adverse event reporting system database” 1 and for sharing their considered perspectives. We appreciate the opportunity to address some of the points raised in their commentary.
First and foremost, we would like to clarify that we did not use the term “inverse causality” in our article. We used the term “inverse association,” fully recognizing that association does not imply causation. Our terminology was chosen to highlight potential relationships that merit further investigation and to generate hypotheses rather than to suggest any direct protective effects. The term inverse signal or inverse association has also been used in previously published studies employing a similar methodology.2–5
We do agree that there are notable limitations inherent in using voluntary reporting databases such as the US Food and Drug Administration Adverse Event Reporting System (FAERS) database, and we have highlighted several of those limitations in our article. We used OpenVigil 2.1 MedDRA-v24 6 (rather than raw FAERS data) which employs data cleaning and pre-processing methods on FAERS data. Yet, we acknowledge that the possibility of duplicates cannot be fully eliminated. Regarding the use of controls, we indeed included non-diabetic medications known for their weight loss effects, whether as a potential side effect or primary indication (such as orlistat, phentermine, bupropion, topiramate, zonisamide, amphetamine, and naltrexone). We, however, agree that incorporating additional sensitivity analyses and controls would enhance the robustness of our findings.
Finally, we concur that complementing pharmacovigilance data with other methodological approaches, such as omics approaches and in vitro or in silico testing, can yield more comprehensive insights. However, this integration was beyond the scope of our current report. Our primary objective was to generate hypotheses based on observed potential associations that could be validated through more rigorous methodologies in future studies.
