We conducted this study in accordance with the principles of the Declaration of Helsinki with the approval of the ethics committee of the Institute of Science Tokyo (M2020-058). The committee approved an opt-out method for obtaining consent.

To investigate the clinical characteristics of patients with AIGA and their responses to steroid therapy, we retrospectively collected data from all patients with AIGA who were diagnosed at the Institute of Science Tokyo (Tokyo Medical and Dental University) Hospital from April 2008 to March 2021. A diagnosis of AIGA was made in accordance with the Japanese guidelines [3].

In patients with urticaria, cholinergic urticaria can be diagnosed if a provocation test with sweating stimuli is positive [19, 20]. For the provocation tests, patients were exposed to heat stress in a sauna at 60°C for 15 min, and patients who had wheals induced were considered to have cholinergic urticaria. However, there were a relatively large number of patients who had episodes of exercise- or heat-induced wheals and/or skin pain even though the provocation test was negative. Therefore, we divided the patients into three groups: those with “urticaria and/or pain episodes” and a positive challenge test, those with “urticaria and/or pain episodes” and a negative challenge test, and those without such episodes (Tables 1–4).

We determined a patient’s response to steroid pulse therapy using the methods we reported in the past [6]. Changes in sweat area on the body surface at 1 month after steroid pulse therapy were assessed by a detailed interview (confirmed in some cases by a subsequent starch–iodine test). With appropriate interviewing, it is possible to estimate changes in sweat area with reasonable accuracy (93% accuracy compared with a subsequent starch–iodine test [6]). Patients were considered responders if they had an increase in sweat area of ≥25% of their body surface area. If the patient received multiple steroid pulse therapies, those who met the criteria after at least one were considered responders.

The determination of relapse after remission with steroid pulse therapy also followed the method we previously reported [6]. Patients who responded to steroid pulse therapy were evaluated for changes in sweat area at each outpatient follow-up visit by a detailed interview (confirmed in some cases by a subsequent starch–iodine test). Patients with a reduction in sweat area of ≥25% were considered to have relapsed. The date of relapse was determined as the date on which the patient noticed a decrease in sweating or, if this was not clear, the date of the outpatient follow-up or starch–iodine test. For patients with multiple relapses, only the first relapse was included in the analysis.

To evaluate the associations between serum CEA or total IgE levels and the treatment response, receiver operating characteristic analysis was used to determine cutoff values for serum CEA and serum total IgE levels. Although there are hospital reference values for serum CEA and serum total IgE levels, optimal cutoff values vary by the disease type, disease activity, and study [21, 22]. Regarding serum CEA levels, we calculated the cutoff values for predicting the response to steroid pulse therapy and for predicting relapse after remission. Regarding serum total IgE levels, we calculated the optimal cutoff for predicting relapse after remission. The optimal cutoffs were calculated using the “perfcurve” function [23] of the “Statistics and Machine Learning Toolbox” in MATLAB (version 2024a; MathWorks, Natick, MA, United States). For simplicity, we used a cutoff value of 1,000 IU/mL of serum total IgE, which has the same sensitivity and specificity as the calculated value of 1,083 IU/mL. The cutoff values for age and “delay” (elapsed time since onset) were the same as those in our previous report [6]. In the Cox regression analysis, the proportional hazards assumption was assessed with Kaplan–Meier curves for the two groups classified according to the defined cutoffs (Supplementary Figure S1).

We performed statistical analyses using the Mann–Whitney U test (Tables 1, 2), Kruskal–Wallis test (Tables 1, 2), chi-square test (Table 3), logistic regression analysis (Supplementary Table S1), log-rank test (Table 4), and Cox regression analysis (Supplementary Table S2). Bonferroni correction was used for multiple comparisons (Table 1). MATLAB was used for data processing, including P-value calculation and visualization.

To investigate the proportion of patients with high serum CEA and/or total IgE levels, we collected data from 210 patients with AIGA. The male/female ratio of the patients was 3.2, and the median age was 27 years [interquartile range (IQR), 19–40], consistent with previous reports [1, 3, 12]. Of the 210 patients, 143 and 180 patients were assessed for serum CEA and total IgE levels, respectively, before steroid pulse therapy. The median serum CEA level was 6.1 ng/mL (IQR, 2.6–11.9 ng/mL). No patients were found to have malignant tumors. The median total IgE level was 75 IU/mL (IQR, 20–294 IU/mL).

We observed an opposite trend between serum CEA and total IgE levels (P = 0.0030, Table 1). When these markers were examined in relation to clinical characteristics, higher serum CEA levels were found in male patients than in female patients and in patients who were examined within 1 year after onset than in those who were examined at ≥ 1 year after onset (P < 0.0001 and P = 0.0010, respectively). Serum CEA levels tended to be higher in patients who had episodes of urticaria and/or dermal pain induced by sweating stimuli but had a negative provocation test than those who did not have episodes of urticaria and/or dermal pain (P = 0.0439). However, higher serum total IgE levels were found in younger patients than in older patients and in patients with coexisting atopic dermatitis than in those without coexisting atopic dermatitis (P = 0.0172 and P < 0.0001, respectively, Table 2).

We examined whether the serum markers had relationships with responsiveness to steroid pulse therapy (Supplementary Figure S2). The response rate to steroid pulse therapy among our patients [125 out of 177 (71%)] was comparable to those in previous reports [1, 6]. We found that patients who responded to the therapy showed higher CEA levels than non-responders (P = 0.0192, Table 1). The optimal serum CEA cutoff level for predicting a treatment response was 1.6 ng/mL (P = 0.0011, Table 3). Other factors, namely, male sex, non-concomitant atopic dermatitis, and <1 year from disease onset to treatment (P = 0.0190, P = 0.0185, and P = 0.0128, respectively), were also associated with a good treatment response (Table 3). The odds ratio for high serum CEA levels after adjustment for these factors was 4.23 (P = 0.0421, Supplementary Table S1).

Next, we explored the correlation between the serum markers and relapse rate after remission achieved by steroid pulse therapy (Supplementary Figure S2). We conducted a time-to-event analysis involving the 125 patients who responded to pulse therapy. The traces were started on the last day of steroid pulse therapy and ended at relapse [44 of 125 (35%)] or were censored if the primary doctor decided to end follow-up [24 of 125(19%)], the patient did not attend the return visit [24 of 125 (19%)], or the end of the study (13 August 2022) was reached [33 of 125 (26%)]. Follow-up to relapse or for at least 12 months was completed for 74% (93 of 125) of patients, and the median follow-up duration for patients with no relapse was 13.5 months. Most relapses were seen within 12 months of remission, with an estimated relapse rate of 38% at month 12 (Supplementary Figure S3).

To examine the effect of serum CEA and total IgE levels on relapse using the time-to-event data, we estimated optimal cutoff values (4.9 ng/mL and 1,000 IU/mL, respectively). In univariate analysis, there was a trend toward more frequent recurrence in the low serum CEA and high serum IgE groups (P = 0.0145 and P = 0.0009, respectively; Table 4; Supplementary Figure S1). We then calculated the hazard ratio incorporating the two serum markers and sex because female patients showed more frequent recurrence than male patients (P = 0.0360). The adjusted hazard ratio was 2.41 for the low CEA group (P = 0.0337) and 4.56 for the high IgE group (P = 0.0083, Supplementary Table S2).

The disease activity of AIGA is affected by the seasons, which may affect the analysis. We previously reported that steroid pulse therapy was effective from March to September [6]. Therefore, in the current study, we compared the number of courses of steroid pulse therapy administered during this period between the low (<1.6 ng/mL) and high (≥1.6 ng/mL) CEA groups, but found no significant difference between the groups (median of 1.5 and 1.0 courses, respectively; Mann–Whitney U test, P = 0.7543). We also reported that relapses were more frequent in the autumn (September to November) than in other seasons, which may bias the relapse rate depending on whether the observation period includes the autumn [6]. Therefore, we reanalyzed the data by excluding cases in which the follow-up period after remission did not include autumn. However, we still found that the relapse rate was higher in the high (≥1000 IU/mL) IgE group than the low (<1000 IU/mL) IgE group (log rank test, P = 0.0010). Therefore, we believe that our results of the analysis of the relationship between serum markers and the effectiveness of steroid pulse therapy were not affected by the seasons.

Finally, we examined whether there was an association between serum markers and the effect of steroid pulse therapy for urticaria and/or dermal pain. Of the 141 patients with urticaria and/or dermal pain, 78 (55%) reported that their symptoms had improved at 1 month after steroid pulse therapy. The low serum CEA (<1.6 ng/mL) and high serum IgE (≥1000 IU/mL) groups of patients had a slightly lower rate of improvement of urticaria and/or dermal pain than the mean rate [4/9 (44%) and 5/13 (38%); chi-square test, P = 0.2908 and P = 0.1536, respectively], but this was not significant. We also examined the rate of re-exacerbation of urticaria and/or dermal pain after steroid pulse therapy. Approximately half of the patients whose symptoms improved experienced re-exacerbation within 12 months (Supplementary Figure S4A). Regarding serum CEA levels, there was no significant difference in the rate of re-exacerbation between the low (<4.9 ng/mL) and high (≥4.9 ng/mL) CEA groups (Supplementary Figure S4B). Regarding serum total IgE levels, only five patients with high serum total IgE levels (≥1000 IU/mL) experienced an improvement of urticaria and/or dermal pain after steroid pulse therapy, which was insufficient to evaluate the rate of re-exacerbation. Overall, the relationship between serum markers and the effectiveness of steroid pulse therapy on urticaria and/or dermal pain was not as clear as that between serum markers and the effectiveness of steroid pulse therapy on sweating.