This trial was conducted between August 2023 and March 2025 at 27 sites in Japan (Japan Registry of Clinical Trials (https://jrct.mhlw.go.jp/), jRCT2031230139) in compliance with the guidelines for Good Clinical Practice and the provisions of the Declaration of Helsinki. At each trial site, the protocol was approved by the relevant institutional review board. All legal guardians (and patients, when possible) provided written informed consent before any trial-related procedures.

This trial consisted of an 8-week, randomized, double-blind, vehicle-controlled period (double-blind [DB] period) and a 52-week, open-label extension period (open-label [OL] period) (Figure 1). At the initiation of the DB period, patients were randomized 2:1 to receive tapinarof cream 0.5% or vehicle cream. A computer-generated randomization was performed with a dynamic allocation method to balance for age category (2–6 years and 7–11 years) and Investigator’s Global Assessment (IGA) score [17, 18]. After completing all assessments at week 8, patients transitioned to the OL period. When patients experienced worsening of AD in the DB period, they could transition to the OL period early at week 4 or later (defined as the early transition visit) at the investigator’s discretion (i.e., no objective criteria were predefined for early transition). In the OL period, all patients received tapinarof cream 0.5%.

Eligible patients were aged 2–11 years and had a clinical diagnosis of AD according to the criteria of the Japanese Dermatological Association [1]. At screening and baseline, patients were required to have an IGA score of 2 (mild), 3 (moderate), or 4 (severe); an Eczema Area and Severity Index (EASI) score [17–20] of ≥5 (excluding the hairy scalp); and affected body surface area (BSA) of 5%–30% (excluding the hairy scalp). Patients were excluded if they had a significant dermatologic or inflammatory condition that would interfere with the trial assessments, had an acute active skin infection, or used AD therapies within the indicated period before baseline (e.g., phototherapy within 4 weeks, topical corticosteroids within 2 weeks or 1 week; Supplementary Table S1).

Legal guardians (and patients, when possible) were instructed to apply a thin layer of trial treatment once daily (QD) to all AD lesions (except for the hairy scalp), including newly appearing lesions. When AD lesions cleared, application to the areas was continued in the DB period but could be interrupted at the investigator’s discretion in the OL period. Therapies that were indicated for AD or that may have been effective in treating AD were prohibited during the trial (Supplementary Table S1). In the OL period, however, these therapies could be used at the investigator’s discretion as rescue therapy when no improvement from baseline was observed in AD lesions.

The assessments of IGA, EASI, and percent of BSA affected were performed by the investigators at each trial visit. The 5-point IGA, ranging from 0 (clear) to 4 (severe), was used. The Patient-Oriented Eczema Measure (POEM) [21, 22] was completed by legal guardians at each trial visit. The pruritus score is a 5-point scale ranging from 0 to 4 [23], with higher score indicating more severe itch, and was completed by legal guardians (or patients, when possible). The daytime and nighttime pruritus scores were recorded in the diary twice daily at bedtime and upon awakening, respectively, in the DB period and were assessed at each trial visit in the OL period.

In the DB period, the primary endpoint was the proportion of patients with ≥75% improvement from baseline in EASI score (EASI-75) at week 8. Key secondary endpoints were the mean percent change from baseline in EASI score and the proportion of patients who achieved IGA treatment success, defined as an IGA score of 0 (clear) or 1 (almost clear) with ≥ 2-grade improvement from baseline at week 8. Other secondary endpoints included the proportion of patients with ≥50% and ≥90% improvement from baseline in EASI score (EASI-50 and EASI-90, respectively), the proportion of patients with an IGA score of 0 or 1, the mean change from baseline in percent of BSA affected, the mean change from baseline in POEM score, and the mean change from baseline in daytime, nighttime, and maximum pruritus scores. The maximum pruritus score on an assessment day is defined as the greater of the daytime and nighttime scores. Across the DB and OL periods (overall trial period), most of the above endpoints were used for long-term efficacy assessments.

Safety assessments included the incidence and severity of adverse events (AEs), clinical laboratory parameters, and vital signs. Trough concentrations of tapinarof were determined in plasma samples collected at weeks 4, 8, 32, and 60.

In the phase 2 trial of tapinarof in Japanese pediatric patients with AD [16], EASI-75 response rate at week 8 was 65.0% in the tapinarof 0.5% group and 15.0% in the vehicle group (patients with missing data were treated as nonresponders). On the basis of this result, a sample size of 63 patients (42 patients in the tapinarof 0.5% group, 21 patients in the vehicle group) would provide at least 90% power to detect a significant difference in the primary endpoint between the groups with a Fisher’s exact test (2-sided p value of <0.05). However, to ensure an adequate number of patients for long-term safety assessments, the target sample size was determined to be 150 patients (100 patients in the tapinarof 0.5% group, 50 patients in the vehicle group).

For the DB period, the primary analyses of efficacy were performed in the full analysis set, which consisted of all randomized patients who received at least 1 application of trial treatment and underwent the assessment of EASI score at least once. For patients who underwent early transition to the OL period, data corresponding to those at week 8 after the transition were handled as missing in the primary and key secondary endpoint analyses of the DB period. EASI-75 response rate at week 8 (the primary endpoint) was analyzed on the basis of 100 datasets where missing data were imputed by the multiple imputation with the fully conditional specification (FCS) model. The FCS model successively imputed missing data from week 2 to week 8, using linear regression models including the treatment group, baseline value, and post baseline values as covariates. The Cochran–Mantel–Haenszel test stratified by baseline IGA score (2 or ≥3) was performed at the 2-sided significance level of 5%. The common between-group difference in the rate and its 95% CI (Sato method [24]) were calculated by combining the between-group differences in each baseline IGA score stratum. The mean percent change from baseline in EASI score at week 8 (a key secondary endpoint) was analyzed with mixed effect models for repeated measures. The mixed effect models for repeated measures included fixed effects for treatment group, analysis visit, treatment group by analysis visit interaction, patient as a random effect, and baseline score as a continuous covariate, with an unstructured variance-covariance structure for the random effects. IGA treatment success at week 8 (the other key secondary endpoint) was analyzed on the basis of the first dataset out of 100 datasets where missing data were imputed by the multiple imputation with the FCS model. Other secondary endpoints were analyzed on the basis of observed cases (OC) where missing data were not imputed. Except for the primary endpoint, no formal statistical tests were performed for comparisons between the groups.

For the overall trial period, the efficacy analyses were performed in the efficacy analysis population, which consisted of all patients who received at least 1 application of tapinarof cream and underwent the efficacy assessment at least once (i.e., vehicle-treated patients who discontinued the trial during the DB period [by week 8] were excluded from the analyses). Efficacy endpoints were analyzed on the basis of OC where missing data were not imputed. Data obtained after initiation of rescue therapy were included in the analyses. For patients who underwent early transition to the OL period, the data at each visit after the transition were assigned according to the specified analysis window.

The safety analyses were performed in the safety analysis population, which consisted of all patients who received at least 1 application of trial treatment for the DB period and who received at least 1 application of tapinarof cream for the overall trial period. The pharmacokinetic analyses were performed in the pharmacokinetic analysis population, which consisted of all patients who had plasma concentration data of tapinarof (including plasma concentration below the lower limit of quantification [LLOQ, 50.0 pg/mL]) at ≥ 1 time point for the overall trial period.

At the initiation of the DB period, 162 patients were randomized to tapinarof (n = 107) or vehicle (n = 55). The 8-week treatment in the DB period was completed by 102 patients in the tapinarof group and 30 patients in the vehicle group. Because of worsening of AD, 2 patients in the tapinarof group and 17 patients in the vehicle group underwent the early transition visit and transitioned to the OL period early, before week 8. A total of 151 patients entered the OL period, and 143 patients completed the trial. The primary reasons for trial discontinuation were AEs and withdrawal by patients (Figure 2).

No apparent differences between treatment groups were noted in the demographics and baseline disease characteristics. At baseline, approximately 60% of patients had a baseline IGA score of 3 (moderate), and few patients had an IGA score of 4 (severe) (2 patients in each group). The mean (SD) baseline EASI score was 10.9 (4.3) in the tapinarof group and 10.7 (3.9) in the vehicle group (Table 1; Supplementary Table S2).

In the OL period, rescue therapy was used at least once by approximately 20% of patients. The rescue therapy used was topical corticosteroids only; no other topical or systemic rescue therapy was performed (Supplementary Table S3). Although treatment with tapinarof could be interrupted at the investigator’s discretion in the OL period after clearance of AD lesions, the duration of exposure to tapinarof was ≥48 weeks in approximately 80% of patients, indicating that most patients continued treatment with tapinarof after the clearance.

In the DB period, significantly more patients achieved the primary endpoint, EASI-75 response rate at week 8, in the tapinarof group (52.59%) than in the vehicle group (12.60%), with a between-group difference of 40.0 percent points (95% CI, 26.3 to 53.7; p < 0.0001) (Figure 3). Greater improvements in the key secondary endpoints were noted in the tapinarof group. At week 8, the least-squares mean percent change from baseline in EASI score was −68.28% in the tapinarof group and −23.01% in the vehicle group, with a between-group difference of −45.27 percent points (95% CI, −59.48 to −31.06) (Figure 4). At week 8, IGA treatment success rate was 26.2% in the tapinarof group and 1.8% in the vehicle group, with a between-group difference of 24.4 percent points (95% CI, 13.2–34.0) (Figure 4). Representative clinical images of patients who achieved IGA treatment success are presented in Supplementary Figure S1. Greater improvements in the tapinarof group were also noted in other secondary endpoints (Supplementary Figure S2), and the maximum pruritus score decreased rapidly in the tapinarof group (Figure 5).

In the tapinarof group, EASI-75 response and IGA treatment success rates continued to increase after week 8. EASI-75 response rate was 53.4% at week 8, 75.8% at week 32, and 80.2% at week 60; IGA treatment success rate was 24.3% at week 8, 48.5% at week 32, and 56.3% at week 60. In the vehicle-to-tapinarof group (patients who received vehicle cream in the DB period), these rates increased rapidly after switching to tapinarof. EASI-75 response rate was 21.3% at week 8, 84.8% at week 32, and 82.8% at week 60; IGA treatment success rate was 4.3% at week 8, 54.3% at week 32, and 62.1% at week 60. In both groups, the increased rates were maintained through week 60 (Figure 6). Similarly, improvements in the other efficacy endpoints were maintained through week 60 (Supplementary Table S4). No apparent differences in efficacy results were noted between the age categories (2–6 years and 7–11 years) (Supplementary Table S5).

In the DB period, AEs were reported in 78 of 107 (72.9%) patients in the tapinarof group and in 41 of 55 (74.5%) patients in the vehicle group (Table 2). No serious or severe AEs were reported in either group. The most common AEs included nasopharyngitis (36.4%), influenza (14.0%), and AD (5.6%, mostly reported as worsening of AD) in the tapinarof group and AD (43.6%), nasopharyngitis (25.5%), influenza (10.9%), and contact dermatitis (5.5%) in the vehicle group. Treatment-related AEs were reported in 6.5% of patients in the tapinarof group and in 3.6% of patients in the vehicle group. The most common treatment-related AEs included application site folliculitis (2.8%) in the tapinarof group and AD (3.6%) in the vehicle group. The incidence of trial discontinuations due to AEs was lower in the tapinarof group (1.9%) than in the vehicle group (14.5%). The most common AE leading to trial discontinuation was AD, which was reported 1.9% of patients in the tapinarof group and 12.7% of patients in the vehicle group.

In the total safety analysis population of the overall trial period, AEs were reported in 150 of 154 (97.4%) patients (Table 2). One serious AE (tibia fracture) and one severe AE (liver function test increased) were reported but were considered unrelated to tapinarof. The majority of AEs were mild or moderate. Common AEs occurring in ≥5% of patients are provided in Supplementary Table S6. The most common AEs included nasopharyngitis (67.5%), influenza (37.7%), AD (16.9%), and skin papilloma (16.2%). Treatment-related AEs were reported in 14.3% of patients. The most common treatment-related AEs included application site folliculitis (4.5%), acne (3.9%), and application site acne and AD (2.6% each). The most common AE leading to trial discontinuation was AD (2.6%).

The incidence of AEs did not increase with continued treatment over 60 weeks (Supplementary Table S7). In the subgroup analysis by age category (2–6 years [younger group] and 7–11 years [older group]) (Supplementary Table S8), the incidence of acne was numerically higher in the older group (15.3%) than in the younger group (4.3%). However, more than half of acne events in the older group were considered unrelated to tapinarof, and there did not appear to be a large difference in the incidence of tapinarof-related acne between the age groups (4.7% in the older group vs. 2.9% in the younger group). No clinically significant changes over time were noted in clinical laboratory parameters or vital signs. Additionally, the plasma concentration of tapinarof was below the LLOQ (50 pg/mL) in ≥80% of patients at each time point (Supplementary Table S9).