Machine Learning Approach Finds Several Severity-Associated Factors in Atopic Dermatitis

Age, male sex, and several other factors may increase risk of severe disease in patients with atopic dermatitis (AD), according to study findings published recently in JAMA Dermatology.

Cited as the most common chronic inflammatory skin disease, AD has been shown to have a high impact on patients’ quality of life, productivity at work, and health care utilization. In understanding its complex and multifactorial pathophysiology, researchers said that there are a wide range of clinical phenotypes in AD, with severity ranging from minimal to erythrodermic forms involving the whole body and many different disease courses.

“Apparently, the different genetic, immunologic, and environmental factors do not contribute in the same extent to every patient,” they noted. “This requires a more detailed definition of the heterogeneous (endo)phenotypes. Owing to recent efforts of a precision medicine approach in AD, there has been a growing interest in deep phenotyping.”

Seeking to identify severity-associated factors in adolescent and adult patients with AD, they performed deep phenotying of cross-sectional data derived from the baseline visit of a prospective longitudinal study called the CK-CARE program. Study participants were inpatients and outpatients with AD from the Department of Dermatology and Allergy of the University Hospital Bonn who were enrolled between November 2016 and February 2020.

In the study, 367 patients (mean [SD] age, 39 [17] years; 157 male [42.8%]; 94% adults) were stratified by severity groups via the Eczema Area and Severity Index (EASI). “The associations of 130 factors with AD severity were analyzed applying a machine learning–gradient boosting approach with cross-validation–based tuning as well as multinomial logistic regression,” researchers explained.

Of the cohort, 177 (48.2%) had mild disease (EASI 7), 120 (32.7%) had moderate disease (EASI > 7 and 21), and 70 (19.1%) had severe disease (EASI >21). Regarding atopic stigmata (minor physical markers of AD), cheilitis (OR, 8.10; 95% CI, 3.35-10.59), white dermographism (OR, 4.42; 95% CI, 1.68-11.64), Hertoghe sign (OR, 2.75; 95% CI, 1.27-5.93), and nipple eczema (OR, 4.97; 95% CI, 1.56-15.78) were all associated with increased probability of severe AD.

Moreover, the probability of severe AD was associated with total serum immunoglobulin E levels greater than 1708 IU/mL and eosinophil values greater than 6.8%, whereas female sex was indicated to reduce probability by 70% (OR, 0.30; 95% CI, 0.13-0.66). Alopecia areata was associated with moderate (OR, 5.23; 95% CI, 1.53-17.88) and severe (OR, 4.67; 95% CI, 1.01-21.56) AD.

Several age-related factors were shown to be associated with AD severity:

• Patients aged 12 to 21 years or older than 52 years had an elevated probability of severe AD
• Patients aged 22 to 51 years had the highest probability of mild AD (peak, age 26 years)
• Age at AD onset older than 12 years was associated with increased probability of severe AD up to a peak at 30 years
• Age at onset older than 33 years was associated with moderate to severe AD
• Childhood onset was associated with mild AD (peak, 7 years)

Lifestyle factors were also shown to be associated with severe AD, including physical activity less than once per week and former smoking.

In examining the predictive performance of machine learning–gradient boosting vs multinomial logistic regression, both approaches were found to differ only slightly (mean multi-class area under the curve value: 0.71 [95% CI, 0.69-0.72] vs 0.68 [95% CI, 0.66-0.70], respectively).

“The associations found in this cross-sectional study among patients with AD might contribute to a deeper disease understanding, closer monitoring of predisposed patients, and personalized prevention and therapy,” they concluded.

Reference

Maintz L, Welchowski T, Herrmann N, et al. Machine learning–based deep phenotyping of atopic dermatitis severity-associated factors in adolescent and adult patients. JAMA Dermatol. Published online November 10, 2021. doi:10.1001/jamadermatol.2021.3668