As of today, the JAK inhibitors approved in Europe for the treatment of RA are tofacitinib, baricitinib, upadacitinib, and filgotinib. These molecules have demonstrated superiority over placebo in treating active rheumatoid arthritis unresponsive to conventional synthetic DMARDs [20]. The JAK-signal transducer and activator of transcription (STAT) pathway is a major downstream intracellular signaling system that plays a crucial role in orchestrating immune responses and controlling hematopoiesis and inflammation. JAKi, with varying selectivity for JAK1, JAK2, JAK3, and Tyk2, have then the role of disrupting proinflammatory cytokine cascades in rheumatic diseases [21]. With the release of the new EULAR guidelines concerning the utilization of JAK inhibitors for RA treatment [14], there is potential for a defensive medicine scenario. This could result in patients not receiving the most appropriate therapeutic option due to the presence of RFs, effectively hindering the achievement of precision medicine. Therefore, there is an increasing need for safety data from both randomized controlled trials (RCTs) and real-world cohorts to better understand the role of individual RFs in the occurrence of AEs during JAKi treatment, especially considering that comparative data from registries and real-world settings regarding SAEs, and neo-plasms, currently conflict with those from the ORAL surveillance study [22,23,24,25,26]. In a “real world” multi-database study by Farzin et al. on RA patients, no evidences emerged for increased risk on CV outcomes with TOFA, when compared to TNFi [22]. Nevertheless, the authors of the paper highlight a higher, though statistically non-significant, risk of CV events associated with TOFA, as opposed to individuals treated with TNFi, among those with existing CV RFs or a history of prior CV events [22]. In a Taiwanese registry of patients with RA, a total of 3179 subjects were examined, including 2357 treated with TNFi and 822 treated with JAKi [23]. Regarding coronary disease, the incidence was 0.48 and 0.45 per 100 patients/year for patients treated with JAKi and TNFi, respectively (p = 0.94). The incidence of stroke was 0.33 and 0.46 per 100/patient years for subjects treated with JAKi and TNFi, respectively (p = 0.55), while the incidence of deep venous thrombosis was 0.26 and 0.44 for patients treated with JAKi and TNFi, respectively (p = 0.3). Finally, the incidence for malignancies was 0.39 in the JAKi group and 0.35 in the TNFi group (p = 0.83). Data from this paper showed similar safety outcomes, risk of AE and mortality in the JAKi group compared with the TNFi group [23]. Accordingly, data from the CORRONA registry showed similar AE (serious infectious events, major adverse CV events, malignancy, death and VTE) rates between RA patients treated with JAKi and various bDMARDs, with the exception of VZV reactivations, which had a significantly higher rate among the JAKi group (HR 2.32; 95% CI) [24]. A Korean study by Cho et al. on 346 RA patients showed a higher frequency of AEs reported among patients treated with JAKi, but the difference was not statistically significant (75 of 196 patients [38.3%] vs. 43 of 150 patients [28.73%], p = 0.105) [25]. Regarding SAEs, there was no intergroup difference in the frequency between subjects treated with JAKi and bDMARDs (4.6% vs. 4.0%, respectively, p = 0.789) [25]. Finally, a retrospective analysis of the Hong Kong Biologics Registry (2471 RA patients, 551 treated with JAKi and 1920 treated with TNFi) did not show an increase of major CV events (incidence 1.34 [JAKi] vs 0.75 [TNFi] per 100 patient-years; p = 0.22) or malignancies (0.81 [JAKi] vs 0.85 [TNFi] per 100 patient-years; p = 0.25) in patients treated with JAKi when compared to TNFi users [26]. Contrarily, in the same cohort there was a higher incidence for non-serious infections (16.3 vs 9.9 per 100 patient-years; p = 0.02) and HZV reactivation (3.49 vs 0.94 per 100 patient-years; p < 0.001) in the JAKi group when compared to TNFi users [26].

In our cohort, most patients (78.6%) presented at least one RF and more than half subjects presented 2 or more RFs, with hypercholesterolemia, smoking habit, age ≥ 65 years and hypertension being the most frequently reported. Regarding AEs, we reported an incidence rate of 28 AEs and 6 SAEs per 100 patients/year, respectively, while the incidence rates of infections, serious infections, and herpesvirus reactivations were 21.6, 3.6, and 8.4 per 100 patients/year, respectively. Moreover, in our cohort, around 12% of patients discontinued the treatment for safety reasons, in line with previous data on BAR from the real-world setting showing a 9.5% frequency of drug discontinuation due to AEs [27]. Concerning serious infections, the incidence rate computed from our data is consistent with what is reported in RCTs [28,29,30], while the incidence rate of herpesvirus reactivation is higher in our cohort than in controlled studies [30, 31]. The last finding could be explained by the real-world setting in which patients do not go through a selection process that often excludes subjects with comorbidities and/or older age. Furthermore, it should be mentioned that, in our cohort, only one patient received anti-VZV vaccination at the start of the therapy, as the recombinant peptide vaccine has only recently become available in the Italian market.

As for thromboembolic events, we observed one case of myocardial infarction in a patient treated with UPA, one case of myocardial infarction MINOCA type in a patient treated with TOFA, and one deep venous thrombotic event in a patient treated with BAR. All patients who experienced a CV event or VTE were over 65 years of age. We calculated a 1.2/100 patients/year incidence rate for VTE, which is consistent with the recently published data from Hong Kong biologics registry, where a 1.34 rate was calculated for CV, cerebrovascular and peripheral vascular events [26]. We could not demonstrate a clear association between the presence of RFs and the occurrence of CV and/or thromboembolic AEs during JAKi therapy. However, given the low number of such events in this cohort, we are unable to draw firm conclusions.

Overall, we did not find a correlation between the frequency of AEs and age ≥ 65 years per se, since in our cohort there were two peaks, one in subjects older than 70 and the other in those in their forties. On the other hand, despite lacking statistical significance, the majority of subjects who experienced AEs had at least one of the RFs studied, although this association had no impact on the frequency of treatment discontinuation. This finding sounds reasonable since most CV RFs—both in RA and the general population—also increase the risk of infection (smoking habit, diabetes and reduced mobility to cite a few). In our cohort, disease activity measured by SDAI, CDAI, and DAS28-ESR was significantly higher both at the baseline and during treatment in subjects who developed AEs than those who didn’t. It is well-known from the literature and the clinical practice that a higher disease activity is associated with a higher incidence of AEs in patients with RA. In a large US national cohort study investigating the risk of serious infections in patients with RA compared to those with non-inflammatory rheumatic diseases, the risk of all serious infections, particularly bacterial, respiratory, sepsis, skin, bone and joint infections were significantly increased in patients with RA, and it was higher in those with higher disease activity [32].

Finally, we did not observe differences in the frequency of AEs, including VZV reactivation, when stratifying the patients by the different JAKi molecules used. This last finding could be influenced by the limited sample size as well as the inhomogeneous distribution of the different molecules in our cohort. With this regard, further studies directly comparing different JAKi would be useful to understand whether differences in selectivity for different JAK could influence the tendency to develop VZV reactivation, which stands out as a class-specific AE.

This study has some limitations, such as the absence of a control group to compare safety data in RA patients treated with different DMARD classes, including bDMARDs. This could have provided a more comprehensive pool of information to utilize in the clinical setting, assisting in the selection of the most suitable, safe and effective therapies from a broad array of options available for each patient. Moreover, the small sample size and the limited observation period may have hindered the detection of less common AEs such as cancer, CV or TVE. While this provides reassurance in some respects, it also prevented us from gaining a more profound understanding of one of the major events of interest.

In conclusion, our data did not reveal any direct correlations between the presence of the examined risk factors, including age ≥ 65 years, and a higher frequency of adverse events (AEs). Conversely, a high disease activity seemed to be related with the occurrence of AEs. Therefore, while using a JAKi, it is crucial to conduct a comprehensive patient evaluation that shouldn’t be solely driven by the presence of RFs but also by the pivotal targets of suppressing systemic inflammation and effectively controlling disease activity.