Proposal of Phenotypic Patterns in Facial Atopic Dermatitis: A Single-Center Retrospective Analysis of 100 Cases

Abstract

Background

Atopic dermatitis (AD) is a chronic inflammatory skin disorder characterized by eczematous lesions and pruritus. Facial and neck involvement in AD can be challenging to differentiate from other facial dermatoses, such as allergic contact dermatitis, seborrheic dermatitis, rosacea, and drug-induced facial dermatitis, complicating treatment strategies.

Objective

This study aimed to investigate the distribution of facial involvement patterns in patients with AD and to explore the differential diagnosis, focusing on head and neck lesions.

Methods

A retrospective study was conducted on 100 patients diagnosed with AD at a single medical center. Facial and neck lesions were categorized into 5 distinct patterns—periorificial, centrofacial, diffuse, patch, and mixed—based on anatomical distribution. The study also evaluated the impact of dupilumab treatment on these patterns.

Results

The most frequently affected anatomical regions were the frontal (68%), buccal (60%), and orbital (57%) areas. The most common facial involvement pattern was periorificial (28%), followed by diffuse (24%) and centrofacial (21%). The dupilumab-treated group showed a higher frequency of the diffuse pattern compared to other treatment groups. Additionally, periorbital hyperpigmentation, Dennie-Morgan folds, and cheilitis were common minor features.

Conclusion

The study identified distinct facial involvement patterns in AD, with centrofacial and periorificial patterns being the most prevalent. These findings underscore the importance of recognizing facial and neck involvement in AD for accurate diagnosis and personalized treatment. Further large-scale studies are needed to validate these results and explore the underlying mechanisms influencing facial involvement.

INTRODUCTION

Atopic dermatitis (AD) is a chronic inflammatory skin disease characterized by eczematous lesions and pruritus. It affects approximately 10%–20% of children and 1%–3% of adults worldwide¹. The distribution of AD lesions varies by region and age, with facial involvement showing age-related differences². Among adults aged 18–59 years, facial involvement occurs in 22.8%, whereas in individuals aged 60 years and older, the rate is 13.2%³.

The clinical manifestations of AD differ by age. In infancy, lesions commonly appear on the cheeks, forehead, and scalp, typically presenting as acute eczematous lesions. In children, lesions frequently affect the antecubital and popliteal fossae and often manifest as xerosis. Adolescents and adults tend to develop more localized eczema, commonly affecting the hands, eyelids, and flexures. Adults may also present with chronic hand eczema or head and neck dermatitis (HND). AD lesions can appear on any site, but the face and neck are recognized as commonly affected areas following the classic distribution³. Facial lesions in AD are characterized by pallor, erythema, orbital darkening, and Dennie-Morgan infraorbital folds. Prominent horizontal neck creases and eczema around the ears are also common features of head and neck involvement.

Facial dermatosis is relatively common in clinical practice. Facial and neck involvement in AD is difficult to differentiate from other facial dermatoses and presents additional therapeutic challenges. Facial dermatoses may result from extrinsic factors such as allergic contact dermatitis (ACD) or from intrinsic causes like AD and seborrheic dermatitis (SD). The differential diagnosis depends on factors such as patient age, sex, medical history, and lesion distribution. For example, ACD typically presents on the eyelids with severe pruritus. Facial dermatosis related to topical corticosteroid withdrawal (TCW) manifests as facial papules. HND triggered by Malassezia overgrowth is characterized by erythematous patches on the central face and chin.

However, no prior studies have compared or classified the distribution of AD lesions by specific facial subregions. Thus, this study aimed to elucidate the clinical distribution of head and neck lesions and categorize facial involvement patterns in patients with AD.

MATERIALS AND METHODS

Patients

A total of 100 patients with AD visited the Dermatology Department of Kyung Hee Medical Center between January 2016 and October 2024. All patients were diagnosed according to the Hanifin and Rajka criteria. Patients with AD without skin lesions on the head or neck were also excluded. We collected and analyzed information on the patients’ sex, age, AD treatment history, and comorbidities. This study is a retrospective chart review and does not include any personally identifiable information of patients; therefore, approval from the Institutional Review Board was not required.

Image analysis

We retrospectively reviewed medical records and photographs of patients with facial or neck involvement. The face and neck were anatomically divided into 14 regions: frontal, supraorbital, orbital, infraorbital, temporal, zygomatic, nasal, labial, mental, buccal, parotid-masseteric, neck, auricular, and peri-auricular. The affected areas were analyzed for each patient based on charts and photo review. Patients were then categorized into one of 5 patterns based on the affected areas: 1) periorificial, 2) centrofacial, 3) diffuse, 4) patch, and 5) mixed (Fig. 1). Patients with a history of dupilumab treatment were assessed based on lesions present prior to its initiation. The photographs were independently assessed by 2 dermatologists (KTE, SMK) in a blinded manner. Further analysis was performed only when the results of both evaluators were identical.

Fig. 1. Facial anatomical units across the face and neck regions.

Definition of facial involvement pattern

Patients were classified into 5 facial involvement patterns: periorificial, centrofacial, diffuse, patch, and mixed. The periorificial pattern is defined by lesions primarily around the eyes, nose, and mouth. In AD, periorbital involvement, including eyelid eczema, is common. The centrofacial pattern involves lesions predominantly on the forehead and cheeks. The diffuse pattern indicates widespread lesions across the entire face. The patch pattern features irregular, randomly distributed lesions, often as discoid eczematous plaques. The mixed pattern combines elements of the 4 patterns described above.

Statistical analysis

Microsoft Excel (Microsoft Corp., Redmond, WA, USA) was used to calculate averages and standard deviations.

RESULTS

General data of the enrolled patients

A total of 100 patients were included in this study, comprising 61 males and 39 females. The mean age was 32.58 years (range: 8–85 years), and the average age of AD onset was 18.43 years (range: 0–83 years).

All patients had moderate-to-severe AD and were receiving systemic treatment. Within the past 3 months, the most commonly prescribed treatment was dupilumab (47 patients), followed by cyclosporine (22 patients) and Janus kinase inhibitors (16 patients). For treatments initiated more than 3 months prior, cyclosporine was the most frequently used (58 patients), followed by methotrexate (5 patients) (Table 1).

Table 1. Demographic and treatment characteristics of patients with AD.

Category	Details	No. of patients
Total patients	Total	100
	Males	61
	Females	39
Age information (yr)	Mean age	32.58 (range: 8–85)
	Average age of onset of AD	18.43 (range: 0–83)
Recent treatments (within 3 mo)	Dupilumab	47
	Cyclosporine	22
	Janus kinase inhibitors	16
Past treatments (more than 3 mo ago)	Cyclosporine	58
	Methotrexate	5

Values are presented as median (minimum-maximum).

AD: atopic dermatitis.

Distribution of anatomical units

The face and neck were divided into 14 anatomical regions to evaluate the frequency of involvement (Fig. 1). The most frequently affected region was the frontal area (68%), followed by the buccal (60%) and orbital (57%) regions.

The orbital region was further subdivided into orbital, supraorbital, and infraorbital areas. Among these, the orbital region had the highest involvement (57%), while the supraorbital and infraorbital regions were involved in 46% and 31% of patients, respectively.

Regions in the lower face, such as the labial and mental areas, showed involvement rates below 30%, indicating lower prevalence compared to upper and midface regions. The auricular and periauricular areas had a combined involvement rate of 15%. The neck was involved in 42% of patients, comparable to midface region involvement (Table 2).

Table 2. Distribution of anatomical units.

Unit	No.	Unit	No.
Frontal (F)	68	Labial (L)	29
Supraorbital (S)	46	Mental (M)	20
Orbital (O)	57	Buccal (B)	60
Infraorbital (I)	31	Parotid-masseteric (Pm)	11
Temporal (T)	32	Neck (Ne)	42
Zygomatic (Z)	40	Auricular (A)	10
Nasal (N)	18	Peri-auricular (Pa)	5

Distribution of facial involvement pattern

Facial lesions were categorized into 5 patterns: periorificial, centrofacial, diffuse, patch, and mixed. The most common pattern was periorificial (28%), followed by diffuse (24%), centrofacial (21%), patch (18%), and mixed (9%) (Fig. 2, row I).

Fig. 2. Phenotypic patterns and distribution in facial atopic dermatitis. (A) Periorificial, (B) centrofacial, (C) diffuse, (D) round patch, (E) mixed.

The mixed pattern, defined as the coexistence of 2 or more patterns, was further broken down into its component patterns. When these were counted separately, the periorificial pattern was the most frequent (30.28%), followed by centrofacial (25.69%), diffuse (23.85%), and patch (20.18%) patterns. These frequencies differed slightly from those observed when using the 5-category classification (Fig. 2, row II).

Facial involvement patterns by treatment

Facial involvement patterns were also analyzed according to treatment. Among patients treated with cyclosporine, Janus kinase inhibitors, and methotrexate, the periorificial pattern was the most common—consistent with the overall trend.

Conversely, among patients treated with dupilumab, the diffuse pattern was the most prevalent, followed by the periorificial pattern. This suggests a distinct facial lesion distribution in the dupilumab group compared to the overall patient cohort (Table 3).

Table 3. Facial involvement patterns in atopic dermatitis according to systemic medication.

Treatment medication	Facial involvement pattern	Frequency
Cyclosporine	Periorificial	23
	Diffuse	21
	Patch	17
	Centrofacial	14
	Mixed	5
Dupilumab	Diffuse	17
	Periorificial	13
	Patch	10
	Mixed	4
	Centrofacial	3
Janus kinase inhibitors	Periorificial	7
	Diffuse	4
	Centrofacial	4
	Patch	3
	Mixed	0
Methotrexate	Periorificial	3
	Patch	3

Minor and associated features of AD

The Hanifin and Rajka diagnostic criteria for AD include various minor features, some of which may appear on the face and neck. These include cheilitis, Dennie-Morgan infraorbital folds, and facial erythema or pallor.

In this study, 20 patients exhibited periorbital hyperpigmentation, 15 showed Dennie-Morgan folds, and 12 had cheilitis. Hand eczema, another common symptom, was observed in 12 patients, while foot eczema was reported in 2.

Hertoghe’s sign, characterized by thinning or loss of the lateral eyebrows and often associated with AD or hypothyroidism, was observed in 27 patients.

Other associated dermatologic conditions included nummular eczema (6 patients), prurigo nodularis (5 patients), and lichen amyloidosis (4 patients).

Allergic conditions frequently associated with AD were also assessed. Allergic rhinitis was present in 22 patients, allergic conjunctivitis in 2, asthma in 3, and chronic urticaria in 3 (Table 4).

Table 4. Clinical features and comorbidities in patients with AD.

Category	Feature/Condition	No. of patients
Minor signs of AD	Hertoghe’s sign	27
	Periorbital hyperpigmentation	20
	Dennie-Morgan folds	15
	Cheilitis	12
Eczema	Hand eczema	12
	Foot eczema	2
Associated dermatological manifestation	Nummular eczema	6
	Prurigo nodularis	5
	Lichen amyloidosis	4
Associated allergic diseases	Allergic rhinitis	22
	Allergic conjunctivitis	2
	Asthma	3
	Chronic urticaria	3

AD: atopic dermatitis.

DISCUSSION

Facial dermatosis is a relatively common dermatologic condition. Involvement of the face and neck in AD is often difficult to differentiate from other facial dermatoses, which poses additional therapeutic challenges. The differential diagnosis for facial AD includes facial ACD, SD, rosacea, TCW, and drug-associated face and neck dermatitis, among others. These conditions differ in terms of facial involvement areas, and their characteristic clinical features provide essential clues for accurate diagnosis.

In this study, we categorized facial AD into 5 distinct patterns based on the distribution of anatomical units through retrospective chart and photo reviews. These clinical patterns displayed distinct presentations, with the frontal and buccal regions being the most commonly affected. Anatomical analysis revealed that the frontal (68%), buccal (60%), and periorbital (57%) regions were most frequently involved. While the anatomical unit-based approach highlighted involvement of the forehead, cheeks, and periorbital region, the overall pattern analysis showed that the periorificial type was most prevalent, followed by the diffuse pattern.

The centrofacial pattern may overlap with rosacea. Rosacea typically presents with erythema localized to the central face, particularly the nose and cheeks, whereas lesions caused by M. furfur overgrowth often involve seborrheic areas like the eyelids, forehead, and chin. Accurate diagnosis may require Demodex testing, and treatment may benefit from tetracycline-class antibiotics. SD, which affects sebaceous-rich regions such as the central face, scalp, and anterior chest, can mimic AD. It often presents in the nasolabial and alar folds and may include blepharitis involving the anterior eyelid margin(lash) line is a frequent finding. When AD affects the cheek, it is particularly important to differentiate it from SD⁴.

The periorificial type may represent an exacerbated form of AD. Eyelid involvement in AD must be carefully differentiated from ACD, which is typically marked by papulovesicles on an erythematous base and intense pruritus. Persistent eyelid dermatitis warrants strong suspicion of ACD⁵. As ACD and AD often co-occur, differentiating between them is particularly challenging in eyelid involvement. For such presentations, patch testing and allergen avoidance are recommended. In some cases, evaluation for photosensitivity may be necessary. In this study, the periorbital pattern was the second most common, suggesting a potential overlap between ACD and AD and reinforcing that eyelid dermatitis may be a characteristic manifestation in patients with AD. Treatment of the periorificial type includes standard oral and topical therapies used for AD.

Patch-type facial lesions identified in this study may be associated with pharmacological agents, particularly dupilumab⁶. Facial and neck erythema related to dupilumab has been increasingly reported, with an incidence of up to 10% in clinical practice⁴. Among patients who developed facial and neck dermatitis while on dupilumab, over half had pre-existing lesions in these areas. However, these new lesions often differ from prior AD manifestations. The exact mechanism underlying dupilumab-associated facial erythema remains unclear. Proposed factors include rosacea, alcohol-induced flushing, ACD, and Malassezia-associated HND. The activation of the Th1/Th17 pathway may promote Malassezia proliferation⁷. These conditions are typically differentiated based on the distribution of lesions in the face and neck region. Therefore, when facial erythema occurs in AD patients, clinicians should assess for dupilumab use.

In this study, patients treated with dupilumab most commonly exhibited the diffuse pattern. This finding may reflect treatment patterns in Korea, where patients receiving dupilumab generally have severe AD, and those referred to tertiary care centers often present with extensive facial involvement, introducing a potential selection bias. Supporting this, a previous study reported that 84.6% of patients exhibited facial dermatitis attributable to AD prior to initiation of dupilumab therapy⁸.

From a mechanistic standpoint, interleukin-4 (IL-4) has been shown to elevate reactive oxygen species, resulting in endothelial nitric oxide synthase inhibition and decreased nitric oxide bioavailability, which may contribute to endothelial dysfunction and vasoconstriction. Accordingly, IL-4 blockade could theoretically exacerbate vascular dilation; however, empirical evidence supporting this hypothesis remains limited, warranting further investigation⁹.

From a pharmacological standpoint, facial dermatosis caused by discontinuation of topical corticosteroids (TCS) is another crucial differential diagnosis, commonly presenting with a diffuse pattern. TCS are widely used in AD management, and many patients rely on mid- to high-potency formulations. TCW is a rebound phenomenon that occurs following cessation of prolonged TCS use. Unlike a typical flare, TCW is more severe and persistent. Its pathogenesis is unknown but is often accompanied by systemic symptoms such as fever and flushing. TCW typically affects the face and neck and manifests in 2 subtypes: papulopustular and erythematoedematous¹⁰. Management of TCW-related dermatitis includes the use of topical calcineurin inhibitors.

This study has several limitations. First, it was conducted at a single center with a relatively small sample size, limiting the generalizability of the findings. Larger multicenter studies are needed to validate these results. Additionally, because this study focused exclusively on AD patients, patch testing was not performed, which limited the ability to differentiate AD from conditions such as ACD.

In conclusion, this study analyzed the predilection sites and distribution patterns of facial lesions in AD (Fig. 2). We found that the centrofacial and periorificial patterns were strongly associated with facial AD. We propose that this pattern-based classification may provide a foundational guide for selecting targeted evaluations and treatments. Future studies exploring aggravating factors and allergy profiles within each pattern may enable more personalized management strategies for patients with AD.