This study was a multicenter retrospective case–control study based on the China Type II Inflammatory Skin Disease Clinical Research and Standardized Diagnosis and Treatment Project (NCT05316805). Adult patients (≥18 years) with PN or AD were recruited from more than 100 hospitals in China between June 2021 and January 2024. To enhance comparability, eligible adult patients with PN and AD were enrolled during the same study period at participating centers using standardized diagnostic criteria and harmonized recruitment and data-collection procedures. PN was diagnosed based on predefined clinical criteria, including chronic pruritus for more than 6 weeks, persistent scratching, and the presence of localized or generalized nodular lesions (16). AD was diagnosed according to the Williams Diagnostic Criteria (17). The study was approved by the Clinical Research Ethics Committee of Peking University First Hospital (Approval No. 2021/223) and the Ethics Committee of Peking University Shenzhen Hospital (Approval No. 2022/022). All participants provided written informed consent at enrollment. Supplementary Figure S1 shows the flowchart of the participant selection process for the study.

At enrollment, sociodemographic data, including age, sex, body mass index (BMI), residence, occupation, education, smoking status, and alcohol use were collected. Additionally, allergic and atopic histories, hematologic and allergy-related biomarkers, and comorbidities were recorded. Overall clinical status and characteristics were documented along with standardized assessments by dermatologists, such as the Investigator's Global Assessment (IGA), peak pruritus Numerical Rating Scale (pp-NRS), Dermatology Life Quality Index (DLQI), and Hospital Anxiety and Depression Scale (HADS).

Four prespecified comparisons were defined: (1) PN vs. AD (Reference: AD); (2) IgE-high PN vs. IgE-normal PN (Reference: IgE-normal PN); (3) IgE-high PN vs. AD (Reference: AD); and (4) IgE-normal PN vs. AD (Reference: AD). PN subgroups were defined based on a site-reported binary indicator of elevated total serum IgE at enrollment, using each center's local laboratory reference range. Absolute IgE values and assay-specific thresholds were not available in the analytic dataset. Therefore, IgE status was treated as a pragmatic indicator rather than a standardized cutoff. Covariates included demographic and lifestyle factors, atopic and clinical allergy history, hematologic and allergy-related biomarkers, and comorbidities. Age was analyzed in categories (18–44, 45–59, 60–74, ≥75 years), and BMI was grouped based on World Health Organization standards.

Baseline characteristics were summarized using descriptive statistics. Multivariable logistic regression was used to estimate adjusted odds ratios (aORs) with 95% confidence intervals (CIs) for the prespecified comparisons. For each binary comparison, the dependent variable was coded with the target group as 1 and the reference group as 0; therefore, an aOR > 1 indicates higher odds of belonging to the target group relative to the stated reference group. Group ordering and labels for each comparison are provided in the figure and table legends. Missing data were handled using multiple imputation (m = 20). Imputed datasets were analyzed separately and pooled using Rubin's rules. Total IgE status was handled differently across analyses. In the primary PN vs. AD comparison, IgE status was treated as a covariate and was included in the imputation procedure when missing. In contrast, in the within-PN analyses in which PN was stratified by IgE status, IgE status defined subgroup membership and was not imputed, and the stratified analyses were therefore restricted to participants with observed IgE status. Multicollinearity was assessed using generalized variance inflation factors, and statistical significance was defined as a two-sided P < 0.05.

To understand how IgE status might alter the clinical profile, the analysis tested for interactions between IgE status and key atopic covariates. Age and BMI were additionally explored with restricted cubic splines in supplementary analyses to visualize nonlinear patterns, while the primary analyses retained categorical specification for ease of clinical interpretation.

Discrimination was evaluated using the area under the receiver operating characteristic (AUC) curve, with internal validation using bootstrapping (1,000 bootstrap samples). Calibration was assessed using calibration intercept and slope and the Brier score. All data processing and analyses were conducted using R statistical software version 4.4.1.

A total of 2,621 participants were enrolled in this study, including 1,462 with PN and 1,159 with AD. The median age was 51.8 years (IQR, 38.1–62.3) for PN, and 42.4 years (IQR, 29.7–60.0) for AD. The proportion of females was similar between the two groups. PN was more prevalent among rural residents and those with lower educational levels, whereas AD was characterized by a higher frequency of atopic features. Disease severity at enrollment was generally moderate in both groups, with IGA=3 being the most frequent category. Compared with AD, PN showed a lower patient-reported symptom burden and less quality-of-life impairment: the median pp-NRS was 6.0 (IQR, 3.0–8.0) in PN and 7.0 (IQR, 5.0–8.0) in AD, and correspondingly, the median DLQI was 9.0 (IQR, 4.0–14.0) and 11.0 (IQR, 6.0–16.0), respectively. The median HADS total score was 10.0 (IQR, 3.0–18.0) for PN and 10.0 (IQR, 3.0–16.0) for AD. Additional details are provided in Table 1. Furthermore, the baseline demographic and clinical characteristics of PN patients with and without a history of AD are presented in Supplementary Table S1.

Compared with AD, PN was associated with middle age and several socio–behavioral factors. Patients aged 45–59 years (vs. 18–44 years) had higher odds of PN [adjusted odds ratio (aOR), 1.56; 95% CI, 1.24–1.96; P < 0.001], as did those residing in rural areas (aOR, 1.67; 95% CI, 1.29–2.18; P < 0.001). Lower educational attainment (primary or below vs. bachelor's or higher) was associated with an aOR of 2.00 (95% CI, 1.41–2.84; P < 0.001), as was current smoking (aOR 1.84; 95% CI, 1.30–2.61; P < 0.001). In contrast, atopic indicators were lower in PN than in AD. These included elevated total serum IgE (aOR 0.73; 95% CI, 0.58–0.91; P = 0.006), peripheral eosinophilia (aOR 0.75; 95% CI, 0.58–0.98; P = 0.033), allergen-specific IgE class ≥2 (aOR 0.43; 95% CI, 0.31–0.60; P < 0.001), the family history of atopic diseases (aOR 0.60; 95% CI, 0.45–0.79; P < 0.001), and allergic rhinitis (aOR 0.47; 95% CI, 0.37–0.60; P < 0.001). Regarding comorbidities, PN was more frequently associated with type 2 diabetes mellitus (aOR 2.13; 95% CI, 1.11–4.09; P = 0.023), and psychiatric disorders were also more frequent (aOR 10.23; 95% CI, 2.46–42.57; P = 0.002). However, coronary heart disease and current alcohol use showed no significant associations. These results are presented in Figure 1.

Further analyses suggested effect modification by IgE status for atopic features in PN. The attenuation of atopic markers observed for PN overall, exemplified by allergen-specific IgE and allergic rhinitis, appeared to be largely driven by the IgE-normal subgroup, whereas the IgE-high subgroup showed an atopic profile more similar to AD. On the multiplicative scale, tests for interaction between IgE status and classical atopic markers were statistically significant for allergen-specific IgE (P for interaction < 0.001), allergic rhinitis (P for interaction < 0.001), and asthma (P for interaction = 0.008). Figure 2 illustrates these stratified associations, and full interaction estimates are provided in Supplementary Table S2.

The relationships between age and BMI and the odds of PN were found to be nonlinear (Figure 3). After adjustment, the odds of PN increased from young adulthood, peaked at approximately 50 years of age, and then attenuated in older patients. For BMI, the odds of PN showed a continuous increase across the observed range (16.5–46.8 kg/m²).

Among 1,462 patients with PN, total serum IgE were available for 1,395 patients (95.4%). Of these, 245 (17.6%) were classified as IgE-high and 1,150 (82.4%) as IgE-normal (Supplementary Table S3). The 67 cases without IgE data were excluded from the IgE-stratified analyses to maintain internal consistency in subgroup comparisons. To assess the potential impact of excluding these cases, Supplementary Table S4 presents a comparison of the baseline demographic characteristics between the 67 patients who lacked IgE measurements and those with available IgE data.

Relative to the IgE-normal PN subgroup, the IgE-high subgroup exhibited a distinctly type 2 immune response-associated inflammatory profile, characterized by peripheral eosinophilia and allergen sensitization. There was a marked increase in peripheral eosinophilia (aOR 4.62; 95% CI, 3.16–6.77; P < 0.001) in the IgE-high subgroup, as well as allergen-specific IgE class ≥ 2 (aOR, 3.55; 95% CI, 1.93–6.50; P < 0.001). Notably, a key feature defining this atopic endotype was its strong association with advanced age: patients ≥75 years were significantly more likely to be in the IgE-high group (aOR, 2.46; 95% CI, 1.23–4.92; P = 0.011). In contrast, there were no significant differences between the subgroups in terms of rural residence and current smoking. These within-PN associations are summarized in Figure 4.

When compared to AD, the IgE-high PN subgroup showed limited differences in most type 2 immunity-related biomarkers. Allergen-specific IgE class ≥2 (aOR 1.00; 95% CI, 0.64–1.57; P = 0.997), allergic rhinitis (aOR 0.98; 95% CI, 0.67–1.42; P = 0.898), and family history of atopic diseases (aOR 0.70; 95% CI, 0.46–1.07; P = 0.102) did not differ significantly between groups. However, peripheral eosinophilia was more prevalent in the IgE-high PN patients (aOR 1.91; 95% CI, 1.37–2.68; P < 0.001). Sociodemographic correlates characteristic of PN, including age, rural residence, and lower educational attainment, persisted within this IgE-high PN subgroup. Full results are shown in Figure 4.

Relative to AD, the IgE-normal PN subgroup demonstrated a distinct non-atopic profile together with consistent socio–behavioral correlates. Rural residence (aOR 1.60; 95% CI, 1.22–2.11; P < 0.001) and lower educational attainment (primary or below; aOR 2.17; 95% CI, 1.48–3.17; P < 0.001) were associated with IgE-normal PN. Current smoking was also significantly associated (aOR 1.64; 95% CI, 1.12–2.40; P = 0.011). Conversely, atopic features were less prevalent than in AD. Allergen-specific IgE class ≥2 (aOR 0.23; 95% CI, 0.15–0.35; P < 0.001) and allergic rhinitis (aOR 0.39; 95% CI, 0.30–0.50; P < 0.001) were both significantly less common in the IgE-normal PN subgroup. These stratified associations are illustrated in Figure 4.

In the primary analysis distinguishing AD from PN, the model achieved acceptable discrimination (optimism-corrected AUC = 0.713) and reasonable calibration in internal validation. External validation is warranted before applying predicted probabilities to other settings. Full performance metrics for all subgroup analyses are provided in Supplementary Table S5.