A Response to: Letter to the Editor Regarding “Geographical Differences in the Safety and Efficacy of Tofacitinib Versus TNFi: A Post Hoc Analysis of ORAL Surveillance”

To the Editor,

We would like to thank Dr. Goyal for their interest in our work that was recently published in Rheumatology and Therapy [1] and the observations made in the Letter to the Editor, which we address below.

Dr. Goyal noted the stratification of patients into three geographical regions in our study, namely Poland, North America, and Other countries. Clinical trials often include small to moderate population samples from various geographical regions and any surprising results observed in a specific country are generally attributed to chance [2]. However, in the ORAL Surveillance study, 759/4362 (17.4%) patients were from Poland [1, 3], which was the country with the largest representation aside from the USA with 1200 (27.5%) patients. In contrast, there were a total of 449 patients from other countries in Eastern Europe (Bulgaria, N = 47; Czech Republic, N = 200; Russia, N = 154; and Slovakia, N = 48) and 116 patients from Western Europe (UK, N = 47; Spain, N = 54; Netherlands, N = 9; and Finland, N = 6). Therefore, analyzing the Polish population provided an opportunity to evaluate safety and efficacy among a large number of patients with similar demographics and healthcare, and served as a comparator for two other regions (North America and Other countries). We opted to include other Central-Eastern European countries in the Other countries category as there could be cultural and healthcare system differences in this region that influence the safety and efficacy of tofacitinib for patients with rheumatoid arthritis. However, it is also likely that as a result of the homogeneity and genetics background of the Polish population [4, 5], the case of Poland may be informative for other Slavic countries. The small sample size of some geographical regions included in the Other countries category, such as Africa (South Africa, N = 278) and Asia (8 countries, N = 232 patients) limited additional analyses and was beyond the scope of this study.

Dr. Goyal also noted the variations in baseline characteristics across regions (including cardiovascular [CV] risk factors and comorbidities) and suggested an adjustment for confounding factors to ensure comparability of results. Our study comprised a post hoc analysis of the ORAL Surveillance trial [3] and investigated the safety and efficacy of tofacitinib vs tumor necrosis factor inhibitors (TNFi) in different geographical regions. Thus, the objective of this post hoc analysis differs from the primary objective of the ORAL Surveillance study (comparing the non-inferiority of tofacitinib vs TNFi in a CV risk-enriched population), for which analyses with statistical models adjusting for the effects of predictors, such as geographic region, were explored and reported [3]. This post hoc analysis summarizes the safety and efficacy outcomes for each treatment group in each classified geographic region and provided supplemental information to the main analyses of the study for the assessment of safety and efficacy endpoints. Therefore, the observed differences in demographics and baseline characteristics, including CV risk factors and comorbidities, across geographical regions are important findings which provide context for the safety differences observed. Specifically, we found that within regions with high proportions of high-risk patients, safety results followed similar trends to the overall population in the tofacitinib and TNFi groups; differences in safety outcomes among different regions were driven by the presence of baseline risk factors.

Regarding differences in healthcare systems and clinical practices, the recognition of such differences in fact prompted the initiation of this post hoc analysis. Any differences in healthcare systems between geographical regions would likely be systematic, affecting all comparator arms to an equivalent extent within a given geographical region, thus not impacting the treatment differences reported within each region.

There was an additional question raised about long-term implications of our findings; while we agree this is an important topic of discussion, this was beyond the scope of our study. The ORAL Surveillance trial followed patients for up to 6 years, with a median follow-up time of 4 years. A thorough discussion of long-term implications beyond this time frame would be speculative.

Finally, we would like to clarify that while this post hoc analysis was funded by Pfizer and included Pfizer employees as authors, it was prompted by Polish clinicians on the basis of their clinical experience and local observations, which suggested some differences in safety findings within the Polish patient population compared with the overall global safety findings of the ORAL Surveillance study. Of note, all adverse events of special interest in ORAL Surveillance were adjudicated by external committees and the trial protocol and statistical analysis plan have been published [3]. We agree with Dr. Goyal that future research to further explore these findings would be welcome.