*3.7. GRADE Approach for CoE*

We presented the CoE using a modified GRADE evidence profile (Table 4) on the GRADEpro website (https://gdt.gradepro.org/ accessed on 23 February 2021). The GRADE guidelines [49] indicated that publication bias levels were downgraded for studies that identified conflicts of interest. The overall CoE of the efficacy outcomes was deemed "moderate." Owing to the "high" bias risk of BREEZE-AD4 2020, the overall CoE of the safety outcomes was considered "low." The NNTs for patients achieving IGA, EASI-75, and pruritus-NRS response were 3.97, 5.72, and 4.91, respectively. The NNH for patients experiencing TEAEs was 14.80.


**Table 2.**Safety outcomes of Janus kinase inhibitors.

 \**p*< 0.05; \*\**p*< 0.01; CI, confidence interval; RR, risk ratio.


#### **Table 3.** Treatment-emergent adverse events reported in the studies included in the meta-analysis.

AD, atopic dermatitis; AE, adverse event; NA, not applicable; RR, relative risk; SAE, serious adverse event; TEAE, treatment-emergent adverse event; URTI, upper respiratory tract infection.


**Table 4.**Certainty of evidence based on GRADE (Janus kinase inhibitors vs. placebo in atopic dermatitis).

 AEs, Adverse Events; CI, confidence interval; GRADE, Grading of Recommendations Assessment, Development and Evaluation; IGA, Investigator's Global Assessment; NNTs/NNHs, number-needed-totreats/number-needed-to-harms; NR: not reasonable (no statistical significance in meta-analysis; therefore, calculation of this value is not reasonable); NRS, Numerical Rating Scale; RCT, randomized controlled trial; TEAEs, treatment emergen<sup>t</sup> AEs. a Potential conflict of interest was indicated. b One study was judged as "high" bias risk.

#### **4. Discussion**

Our study provides evidence that JAK inhibitors are more effective in achieving EASI-75, IGA, and pruritus-NRS responses than placebo in patients with AD. AD patients treated with JAK inhibitors had relatively higher risks of developing TEAEs, but no higher risks were observed regarding AEs leading to drug discontinuation. From the viewpoint of evidence-based medicine based on the GRADE approach, the overall CoE of the efficacy outcomes was "moderate," while that of the safety outcomes was considered "low".

A previous meta-analysis of five RCTs demonstrated that JAK inhibitors lowered EASI and pruritus scores [20], which is consistent with the findings of this study. However, in contrast to the former study with a small sample size that represented only four countries, we included 15 multi-center RCTs in more than 10 countries and performed subgroup analyses and meta-regressions to identify potential effect modifiers of the efficacy and safety outcomes. Additionally, sensitivity analyses confirmed the robustness of our results. We also applied a GRADE assessment and furnished the NNT and NNH results to facilitate evidence-based decisions in the treatment of AD in clinical practice. Hence, our updated meta-analysis is robust and provides more evidence than a previously published metaanalysis [20].

Given that different kinds of JAK inhibitors seemed to have different effects in AD patients, there were no head-to-head comparisons between JAK inhibitors. According to our subgroup analyses of different JAK inhibitors at different time points, gusacitinib was unlikely to achieve EASI-75 and IGA responses. In the original article, gusacitinib showed benefits in achieving EASI-50 but not EASI-75 [47]. This may also be subject to the lack of statistical power due to an insufficient number of participants. Additionally, topical delgocitinib had higher rates of achieving EASI-75 response than placebo but not IGA response. Other topical agents such as tofacitinib or ruxolitinib demonstrated better response than placebos regarding IGA and pruritus-NRS responses. As seen in Table 3, ruxolitinib and delgocitinib seemed to have fewer TEAEs than other JAK inhibitors. This discovery is important to clinicians because these topical agents may serve as better options than oral forms for their effectiveness and fewer side effects.

We selected EASI as an efficacy outcome because it is adequately validated and recommended for the evaluation of the clinical signs of AD in RCTs [50,51]. However, an EASI assessment is not always feasible in routine practice as it is complex and timeconsuming [50,51]. IGA is a rapid and easily interpreted alternative to EASI, and it should be included in the measurement outcomes for clinical trial approval under US drug regulations [52]. Despite the lack of standardization and validation, our meta-analysis of the IGA responses indicated less heterogeneity. Moreover, because EASI and IGA were evaluated by medical professionals, a patient-oriented scale is also critical for the assessment of drug efficacy. Consequently, considering chronic and intense pruritus is the major symptom observed in AD patients, we analyzed the previously validated pruritus-NRS as an efficacy outcome [53].

We performed several subgroup analyses and observed that children and adolescents had a higher RR of attaining an EASI-75 response. Only one RCT exclusively assessed pediatric patients. Therefore, no conclusions could be drawn regarding the efficacy of JAK inhibitors in this population.

The administration of JAK inhibitors was associated with an elevated risk of TEAEs. Nevertheless, most TEAEs were mild and tolerable. Nasopharyngitis, headache, and upper respiratory tract infection were the most common TEAEs observed in the enrolled RCTs, consistent with the findings of other systemic immunomodulators in AD management [54,55]. Recently, the warning black box issued by the Food and Drug Administration (FDA) reported an increased risk of serious cardiovascular problems with an oral form of tofacitinib, a JAK inhibitor in treating rheumatoid arthritis (RA) and ulcerative colitis [56]. Among the management of RA patients, JAK inhibitors (baricitinib, filgotinib, and tofacitinib) have also received warnings from the FDA about the increased risk of thromboembolic events and the higher rates of all-cause mortality. Even though the pharmacological features of the topical tofacitinib in AD may differ from that of oral usage in the above circumstances, these safety issues should not be ignored.

Recent studies have reported that AD exhibits complex dysregulations and multiple clinical phenotypes [57–59]. By contrast, between-study heterogeneity with seven distinct JAK inhibitors was observed to be unremarkable in most analyses. We hypothesized that the intracellular blockade by JAK inhibitors results in relatively less interference with the extracellular environment [60,61] and is indicated by the marked homogeneity among various types of JAK inhibitors in clinical settings. Furthermore, we identified administration route, severity, and treatment duration as potential effect modifiers for TEAE outcomes (Table 2). Systemic drug absorption via oral administration, severe inflammatory reactions in patients with moderate to severe AD, and longer treatment durations over 12 weeks could explain the results we obtained from the meta-regression analyses.

NNT/NNH is the average number of patients undergoing treatment with a particular therapy to achieve one additional positive/negative outcome compared with the placebo [62]. NNT < 10 and NNH ≥ 10 indicate "clinically desirable" benefit or harm of a particular therapeutic intervention [63]. In this study, the NNT was <10 for all efficacy outcomes, representing desirable effects compared with placebos. Despite the NNH for TEAEs being 14.80, the AEs were relatively innocuous. We believe that our findings could be helpful for clinical dermatologists in treating patients with AD.

A key strength of our study is the updated literature review and meta-analysis via an evidence-based approach. We provide CoE based on the GRADE system and calculate the NNTs/NNHs, which could guide clinicians in decision-making for AD treatment. However, the findings of this study must be considered with certain limitations in mind. First, we only compared the effects of JAK inhibitors against placebos. Comparisons of different types of JAK inhibitors were not performed in this study. Accordingly, we assume rigorous head-to-head RCTs to be beneficial in the comparison of the efficacy and safety outcomes of various JAK inhibitors. Second, we could not draw a firm conclusion concerning the efficacy of JAK inhibitors for the treatment of pediatric patients with AD because only one RCT that assessed patients aged less than 18 years was enrolled in this study. Given that pediatric AD is common, additional trials to clarify the effectiveness and safe dosage of JAK inhibitors in children and adolescents with AD are required. Third, it has only been five years since JAK inhibitors were approved for the treatment of AD; consequently, only a few RCTs with long-term follow-ups are available at present. Because we could only elucidate the short-term effects of JAK inhibitors, the results of this meta-analysis do not guarantee the long-term safety of JAK inhibitors. Hence, the evidence-based results presented herein must be interpreted with caution.

#### **5. Conclusions**

This systematic review and meta-analysis provide updated evidence for current AD guidelines. The findings demonstrate that JAK inhibitors have favorable efficacy (overall CoE: "moderate") in the treatment of AD with tolerable safety issues (overall CoE: "low"). However, planned prospective studies involving long-term follow-up of AEs and costeffective analyses could aid clinical decisions in the application of JAK inhibitors for the treatment of AD.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/article/10 .3390/jpm11040279/s1, Figure S1: Flowchart of the Material and Methods: PRISMA flow diagram of the study. RCT, randomized controlled trial, Figure S2: Summary of Risk of Bias Assessment, Figure S3: Funnel Plot of Pruritus-NRS Response. Table S1: Search strategies modified in MEDLINE (a), Embase (b), Cochrane CENTRAL (c), and Web of Science (d), Table S2: Subgroup analyses and meta-regressions of efficacy outcomes, Table S3: Subgroup analyses of efficacy outcomes for various Janus kinase inhibitors at different time points, Table S4: Sensitivity analysis of overall effects of each outcome before and after modified Hartung–Knapp–Sidik–Jonkman (HKSJ) adjustment.

**Author Contributions:** H.-R.T., J.-W.L., L.-Y.C., and T.-L.C. had access to all the study data and take responsibility for the integrity of the data and the accuracy of data analysis. T.-L.C. conceptualized, designed, and supervised the study. H.-R.T., J.-W.L., and T.-L.C. performed data acquisition, analysis, or interpretation. All the authors contributed in drafting the manuscript and provided administrative, technical, or material support. T.-L.C. performed critical revision of the manuscript for important intellectual content. H.-R.T., J.-W.L., and T.-L.C. performed the statistical analysis. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** Ethical review and approval were waived for this study because the data were retrieved from published clinical trials in the public databases.

**Informed Consent Statement:** Patient consent was waived due to because the data were retrieved from published clinical trials, in which informed consent was obtained by the primary investigators.

**Data Availability Statement:** The data in this study were collated from published clinical trials, which could be accessed by the public.

**Acknowledgments:** We thank Ching-Ju Fang (Medical Library, National Cheng Kung University, Tainan, Taiwan) for her assistance and helpful comments that substantially improved the quality of our literature search. We also thank the Department of Medical Research of Hualien Tzu Chi Hospital and the Buddhist Tzu Chi Medical Foundation for their invaluable contributions to the methodology used in this systematic review and meta-analysis.

**Conflicts of Interest:** The authors declare no conflict of interest.
