Unmasking contact sensitization in rosacea: Evidence from a systematic review and meta‐analysis

Dear Editor,

Rosacea is a common chronic inflammatory skin disease with multifactorial pathogenesis. Genetic susceptibility, innate and adaptive immune system dysregulation, neurogenic inflammation, neurovascular changes, microbial dysbiosis, increased colonization by Demodex mites, skin barrier function and multiple aggravating factors have been incriminated. ¹ , ² Clinicopathogenic links between contact dermatitis (CD) and rosacea are possible but not well‐established. Sensitive, inflamed skin may be more susceptible to contact sensitization (CS), and facial CD may trigger or worsen rosacea. Moreover, rosacea patients are exposed to several topical products.

This systematic review and meta‐analysis, following PRISMA guidelines (PROSPERO registration number: CRD42024558654), aimed to determine the prevalence of CS among rosacea patients. A comprehensive search of MEDLINE, EMBASE, and Google Scholar was conducted up to May 2025. Observational studies in adults with rosacea reporting patch‐test results were included; paediatric, non‐human and single‐case reports were excluded. Two reviewers independently extracted data. Risk of bias was assessed by ROBINS‐I and JBI tools, and evidence certainty was classified via GRADE. Pooled odds ratios (ORs, 95% CI) were calculated using random‐effects models (DerSimonian‐Laird); heterogeneity was evaluated by I ² and τ. ² Sensitivity, subgroup and meta‐regression analyses explored sources of variability.

Sixteen studies (2045 patients) fulfilled the inclusion criteria (Table 1). In selected patients (referred for patch testing or with suspected allergic CD) tested with standard series and sometimes with additional haptens/products, the mean sensitization rate was 48.8%; in unselected patients tested with standard series, it was 31.1%, up to 65% when additional substances were included.

TABLE 1.

Studies concerning contact dermatitis in rosacea patients.

First Author, Year (Country)ref	Study type	Study length (years)	Patients (N)	Females (%)	Age (mean ± SD or median; years)	Patient selection a	Rosacea subtypes (%)	Patch test series	Patch test exposure (days)	Reading days	Positive patch test rate (n)	Clinical relevance and relationship with rosacea subtypes
Corazza, 1997 (Italy) 10.1111/j.1600‐0536.1997.tb00379.x	P, SC	3.2	29	86%	53 (mean)	Selected	ETR 52%; PPR 41%; PHR 7%	GIRDCA standard series; preservatives, emulsifiers, perfumes series; patients’ own products	2	2, 3	41.4% (12)	Clinical relevance in 42% of positive cases
Jappe, 2005 (Germany, Austria, Switzerland) 10.1111/j.0105‐1873.2005.00514.x	R, MC	8	361	72%	NR (81% aged ≥40 years)	Selected	NR	DKG standard series (Hermal®); constituents of cosmetics and topical drugs; patients’ own products (118/361 patients)	1–2	3	38.2% (138) with commercial test substances
Çakmak, 2007 (Turkey) http://www.jtad.org/2007/2/jtad71201a.pdf	P, C, SC	NR	25	72%	51.4 ± 2.3	Selected	NR	ESS; cosmetic series (Chemotechnique Diagnostics®)	NR	2, 4, 7	32% (8) with ESS; 32% (8) with cosmetic series
Jappe, 2008 (Germany) 10.1111/j.1468‐3083.2008.02778.x	P, SC	3	78	49%	60 (median)	Unselected	NR	DKG standard series (Hermal®); constituents of topical formulations; preservatives; fragrances; topically applied drugs; patients’ own products (21/78 patients)	2	3	62.8% (49) with commercial test substances; 43.6% (34) with standard series
Pónyai, 2012 (Hungary) https://doi.org/10.1111/j.1468‐3083.2011.04260.x	P, SC	6.5	82	83%	54.4 (mean)	Unselected	NR	Brial GmbH D‐Greven series	2	2, 3, 4, 7	35.4% (29)
Darlenski, 2013 (Bulgaria) 10.1111/exd.12251	P, C, SC	NR	28	NR	NR	Probably unselected	ETR 100%	ESS	NR	NR	14.3% (4)
Diczig, 2018 (Hungary) 10.1111/jdv.14922	R, SC	NR	143	NR	NR	Selected	NR	NR	NR	NR	37.8% (54)
Erdogan, 2018 (Turkey) 10.5021/ad.2018.30.3.290	P, C, SC	NR	65	92%	42.5 ± 12.0	Probably unselected	ETR 32.3%; PPR 64.6%; PHR 3.1%; OI 32.3%	ESS (30 haptens); cosmetic series (57 haptens) (Chemotechnique Diagnostics®)	2	2, 3, 4, 7	38.5% (25); 32.3% with ESS; 30.8% with cosmetic series	No significant correlations between positive reactions and rosacea subtypes
Çifci, 2019 (Turkey) 10.2174/1871530319666190101120437	P, C, SC	NR	40	85%	40.0 ± 12.7	Unselected	NR	Nickel units of the ESS; clobetasol‐17‐propionate	2	2, 3, 4, 7	52.5% (21)
Lee, 2019 (South Korea) Korean J. Dermatol 2019;57(3):117‐25	P, SC	2	40	97.5%	42.3 ± 11.7	Unselected	ETR 87.5%; PPR 67.5%; PHR 5%; OI 5%	Korean standard series (Chemotechnique Diagnostics®); TRUE® (SmartPractice, Mekos Labs); fragrance (26 haptens) and cosmetic (31 haptens) series (TROLAB)	2	2, 4	65% (26)	No significant correlations between positive reactions and rosacea subtypes
Ozbagcivan, 2020 (Turkey) 10.1111/jocd.12989	P, C, SC	3	103	66%	42.3 (mean)	Unselected	ETR 69.9%; PPR 27.2%; PHR 2.9%	European cosmetic series (49 haptens)	2	2, 3, 4, 7	60.2% (62)	No significant correlations between positive reactions and rosacea subtypes; clinical relevance in 35.5% of positive cases
Chen, 2021 (China) 10.1111/jocd.13902	P, C, SC	1	267	100%	G1: 37.5 ± 16.6; G2: 38.2 ± 15.4	Selected	G1: ETR 51.6%; PPR 32.8%; PHR 15.6%; OI 40.2%; G2: ETR 49.6%; PPR 35.9%; PHR 14.5%; OI 19.3%	Cosmetic C‐1000 series (20 haptens) (Chemotechnique Diagnostics®)	2	2, 3, 7	57.3% (153)	No significant correlations between positive reactions and rosacea subtypes
Kim, 2022 (South Korea) 10.1684/ejd.2022.4358	R, SC	4	189	86%	43.0 (median)	Selected	ETR 46.6%; PPR 71.4%; PHR 2.1%; OI 1.6%; GR 1.6%	Korean standard series (KOR‐1000, Chemotechnique Diagnostics®); patients’ own products	2	2, 4	73% (138)	Positive association between PPR and nickel sulphate sensitization; negative association between PPR and sensitisation to balsam of Peru and polysensitization
Magdaleno‐Tapial, 2022 (Spain) 10.1016/j.ad.2022.02.026	R, SC	28	200	81%	44.7 (mean)	Selected	NR	GEIDAC standard series; additional series based on clinical suspicion (preservatives, cosmetics, photopatch testing) (Marti Tor®)	2	2, 4	46.5% (93)	Clinical relevance in 32.25% of positive cases
Heo, 2023 (South Korea) Korean J. Dermatol 2023;61(10):602‐11	R, SC	10.5	345	81%	46.4 ± 13.8	Selected	ETR 57.7%; PPR 41.7%; PHR 0.3%; OI 0.3%	Korean standard series (Chemotechnique Diagnostics®); TRUE® (Mekos Labs)	NR	4	72.5% (250)	Significantly higher positivity rates for cobalt chloride, thimerosal, quinoline mix, and captan in PPR compared to ETR
Alevizou, 2024 (Greece) 10.1159/000536246	P, C, SC	NR	50	80%	51.2 (mean)	Probably unselected	ETR 44%; PPR 34%; PHR 20%; GR 2%; OI 42%	ESS (Chemotechnique Diagnostics®)	2	2, 4, 7	30% (15)

Note: When subtype percentages exceed 100%, overlap among clinical subtypes is present. p‐values were omitted from the table but discussed in the text where relevant.

Abbreviations: C, Controlled study; DKG, German Contact Dermatitis Research Group; ETR, Erythematotelangiectatic rosacea; ESS, European Standard Series/European Baseline Series; G1, 122 patients with suspected allergic contact dermatitis to cosmetics; G2, 145 patients without suspected allergic contact dermatitis to cosmetics; GEIDAC, Spanish Contact Dermatitis and Skin Allergy Research Group; GIRDCA, Italian Group for Research on Contact Dermatitis (currently named SIDAPA); GR, granulomatous rosacea; IVDK, Information Network of Departments of Dermatology; MC, Multicenter study; NR, Not Reported; OI, ocular involvement; P, Prospective study; PHR, Phymatous rosacea; PPR, Papulopustular rosacea; R, Retrospective study; SC, Single‐center study; SD, Standard Deviation.

^a Studies were categorized based on patient selection into two groups: (a) selected/probably selected patients (referred for patch testing due to suspected allergic contact dermatitis or intolerance to topical products); (b) unselected/probably unselected patients (tested regardless of suspected allergic contact dermatitis). The adverb “probably” was added when patient selection could be reasonably inferred from study methods but not clearly stated.

Across studies, the most frequent allergens were nickel (22.1%), cobalt (8.4%), methylisothiazolinone (7.8%), fragrance mix I (7.4%), balsam of Peru (7.3%), Kathon CG (6.4%), thimerosal (6.4%), potassium dichromate (6.3%), formaldehyde (6.1%), octyl gallate (5.4%) and propolis (5%). Sensitization to some allergens was more prevalent among selected patients as compared to unselected patients (e.g. cobalt 11.8% vs. 4.1%; methylisothiazolinone 9.8% vs. 1.5%; fragrance mix I 9.7% vs. 3.7%; Kathon CG 9.1% vs. 1.9%). A focussed evaluation of nickel sensitization revealed a CS rate of 52.5% in rosacea patients and 22.5% in controls (p = 0.006). ³ However, this trend for higher rates of nickel sensitivity was not confirmed by other studies. ⁴ , ⁵ , ⁶

Six controlled studies (538 rosacea patients, 366 controls) were suitable for a meta‐analysis (Figure 1). ⁴ , ⁵ , ⁶ , ⁷ , ⁸ , ⁹ The OR was 2.12 (95% CI: 1.20–3.75; p = 0.0097), increasing to 2.66 (1.65–4.29) after sensitivity analysis (Figure 1a,b). Subgroup results (Figure 1c) showed a stronger association for cosmetic‐series allergens (OR 4.16 [2.91–5.94], I ² = 0%) than for standard series (OR 1.66 [0.95–2.87], I ² = 0%; p = 0.006 between subgroups). No meaningful differences were found when stratifying by selected versus unselected patient groups.

FIGURE 1.

Meta‐analysis of contact sensitization in rosacea. (a) Pooled OR for contact sensitization in rosacea versus controls. (b) Sensitivity analysis after removal of individual studies. (c) Subgroup analysis comparing cosmetic series versus standard series.

Formal statistical testing for publication bias was not performed because fewer than 10 studies were eligible for meta‐analysis, in line with Cochrane recommendations. Nevertheless, visual inspection of funnel plots (data not shown) did not suggest bias.

Meta‐regression identified no significant moderators. The proportion of female participants showed no correlation with CS risk, likely due to the uniformly female‐predominant samples rather than the absence of a sex effect. Publication year revealed no temporal trend. Papulopustular rosacea correlated inversely, while erythematotelangiectatic rosacea showed a positive trend, although neither reached statistical significance. Ocular involvement was non‐contributory. Risk of bias was low‐to‐moderate, and evidence certainty was moderate.

Several limitations should be acknowledged. The included studies varied in design, patient selection, allergen panels and reporting practices. Most were single‐centre studies with modest samples, and few included controls. Moreover, clinical relevance was rarely evaluated (Table 1).

Nevertheless, CS appears common in rosacea and may contribute to disease expression, but larger prospective studies involving unselected patients and standardized patch test protocols are needed, particularly addressing sensitization to cosmetic series haptens and patients' own products.

FUNDING INFORMATION

None.

CONFLICT OF INTEREST STATEMENT

None declared.

ETHICAL APPROVAL

Not applicable.

ETHICS STATEMENT

Not applicable.

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.