Loss-of-function mutations in the gene encoding filaggrin are not strongly associated with chronic actinic dermatitis

Chronic actinic dermatitis (CAD) is an uncommon photosensitivity disease, characterized by a photo-exposed site dermatitis, which may be severe and disabling (Kerr and Ibbotson, 2006) (Supplementary Figure S1). Objective evidence of broadband photosensitivity to ultraviolet B (UVB) and often also UVA and visible wavebands is evident and most patients also have multiple contact allergies (Frain-Bell et al., 1974). The condition occurs worldwide, although is commoner in temperate climates and typically develops in middle aged to elderly males. Most individuals have a history of a preceding dermatitis, either atopic, contact allergic or seborrhoeic, although it can occur de novo (Frain-Bell et al, 1974; Hawk, 2004; Kerr and Ibbotson, 2006). Resolution of photosensitivity is reported in up to 50% of individuals after 15 years or more, with contact allergies persisting (Dawe et al., 2000).

The pathogenesis of CAD is incompletely understood; cases are sporadic and genetic risk factors have not been identified. Histology and immunohistochemistry of ultraviolet radiation (UVR)-induced lesional skin are consistent with an immunological delayed-type hypersensitivity reaction (Menage et al., 1996). Given the associations with contact allergies and atopic eczema, we considered that epidermal barrier dysfunction might be a key factor in pathogenesis. The characteristic photosensitivity indicates aberrant cutaneous responses to UVR are also likely to be implicated. Filaggrin is an important structural protein in the epidermis (Smith et al., 2006). Loss-of-function mutations in the gene encoding filaggrin (FLG) are associated with skin barrier dysfunction and increased risk of atopic eczema (Baurecht et al., 2007; Palmer et al., 2006). Filaggrin is degraded to release amino acids in the stratum corneum and FLG mutations lead to reduction in photo-absorbing amino acids, specifically urocanic acid, a chromophore for UVB which may protect against photo-damage (Barresi et al., 2011; de Fine Olivarius et al., 1996; Mildner et al., 2010). We aimed to test the hypothesis that FLG loss-of-function mutations are associated with increased risk of CAD, supporting a role for lack of intrinsic photoprotection and/or skin barrier dysfunction in the pathogenesis of CAD.

This study was conducted in accordance with the Declaration of Helsinki; written informed consent was obtained from all participants. Sequential patients aged ≥18 years attending the National Photodiagnostic Service and diagnosed with CAD were invited to participate. The diagnosis was based on clinical and photobiological assessment (including monochromator phototesting) by a photodermatologist. Patient-reported history of atopic disease was recorded. Patch testing was performed as clinically indicated. Blood was taken for total IgE levels and DNA analysis. DNA samples were analyzed for FLG mutations within the CAD cases of white European ethnicity and in 100 Scottish population-matched controls. The four most prevalent FLG loss of function mutations in this population were screened (R501X, 2282del4, R2247X and S3247X) as previously described (Kezic et al., 2011). Chi-square and logistic regression analysis were used to test for association between CAD and FLG combined null genotype in Stata® (version 13; StataCorp LP, College Station, Texas, USA).

Fifty-five patients were recruited; clinical characteristics and investigation findings are summarized in Table 1. The mean age of onset of photosensitivity was 50.2 (range 21-77) years. The majority (35/55, 63.6%) had marked UVB sensitivity (below 50% of lower limit of normal) and 25 patients (45.5%) had abnormal photosensitivity to UVB, UVA and visible wavebands (Table 1). Photosensitivity had resolved by the time of recruitment in three cases (5.5%). Most patients (42/55; 76.4%) had positive patch testing, including 37 (88.1%) with multiple allergen reactions. Of those with contact allergy the median number of allergens was 4 (range 2-31). A diagnosis of atopic disease (eczema, asthma, allergic rhinitis, or type I hypersensitivity reaction) was noted in 23 (41.8%) patients. Total IgE levels were elevated in the majority (Table 1).

Table 1. Demographic, clinical and investigative characteristics of CAD cases and population controls.

Characteristic	n (%)
Cases	Total 55
Male 28	(50.9)
Female 27	(49.1)
Fitzpatrick skin type I-III (including 10 FLG heterozygotes)	49 (89.1)
Fitzpatrick skin type IV-VI (FLG genotype undefined)	6 (10.9)
Abnormal photosensitivity by waveband (at time of recruitment)	55 cases
UVB only (305±5nm)	0
UVA only (335/365±30nm)	1 (1.8)
Visible only (400±30nm)	0
UVA and UVB	22 (40.0)
UVA and visible	4 (7.2)
All wavebands (UVB, UVA and visible light)	25 (45.5)
Photosensitivity resolved	3 (5.5)
Degree of UVB (305±5nm) sensitivity*
<10% of normal	17 (30.9)
11-19% of normal	7 (12.7)
20-49%	11 (20.0)
50-75%	11 (20.0)
≥76%	9 (16.4)
Atopic disease
Eczema, asthma, hay fever or food allergy	23 (41.8)
Immunoglobulin E levels (Ku/L)	13 (23.6)
Normal (<100)
100-999	17 (30.9)
1000-5000	10 (18.2)
> 5000	6 (10.9)
Unavailable	9 (16.4)
Positive reactions to contact allergens	42/55 cases (76.4%)
Single	5 (11.9)
Multiple	37 (88.1)
Number of patch test series with positive reaction per patient
Single series	19 (45.2)
2 series	15 (35.7)
3 series	7 (16.7)
4 series	1 (2.4)
Commonest positive patch test series
BCD Standard	38 (90.5)
Plants	12 (28.6)
Sunscreens**	7 (16.7)
Corticosteroids	3 (7.1)
Medicaments	3 (7.1)
Controls	Total 100
Male	50 (50.0)
Female	50 (50.0)

^*Abnormal UVB sensitivity (assessed by monochromator phototesting at 305±5nm (half-maximum bandwidth)): The minimal erythemal dose [MED] expressed as a percentage of the lowest normal value for MED at this waveband in healthy volunteers of skin phototype I-III. The lower the MED, the higher the level of abnormal photosensitivity. Thus, an MED of <60; 10% of normal is the most abnormally photosensitive;

^**sunscreen patch testing does not represent photopatch testing.

Genotyping results were obtained for all four of the FLG mutations in 48/49 cases (Table 2 and Supplementary Figure S2) of whom 10 had a FLG mutation (20.4%). 22/49 (44.9%) of these patients demonstrated atopic co-morbidity and of the cases carrying a FLG mutation, 6/10 had a history of atopic eczema; in one the atopic status was not recorded. In the control group 12/100 (12.0%) individuals were heterozygous for a FLG loss-of-function mutation but this was not significantly different from the prevalence seen in the cases (chi square p=0.16, odds ratio=1.92, 95% confidence interval 0.77 to 4.85). The apparently higher prevalence of FLG mutations within the cases compared to controls may reflect the co-association of atopic eczema with CAD and sub-group analysis of the small number of atopic CAD cases (n=22) showed a borderline but non-significant association with FLG null genotype (p=0.05, OR=2.93, 95% CI=0.95-9.01). In contrast the non-atopic cases (n=26) showed no association with FLG genotype (p=0.95, OR 0.96, 95% CI=0.25-3.67).

Table 2. FLG genotyping results for patients of white European ethnicity.

	Cases n (%)	Controls n (%)
Co-morbid atopic disease	22 (44.9)	Undetermined
FLG wild type genotype	38 (77.6)	88 (88.0)
FLG heterozygote genotype	10 (20.4)	12 (12.0)
FLG homozygote or compound heterozygote	0	0
Incomplete genotyping data0	1 (2.0)	0
Total	49 (100.0)	100 (100.0)
Chi2 analysis	p=0.16
Odds ratio (95% CI)	1.02 (0.77 to 4.85)

Cases with Fitzpatrick skin type of IV-V were excluded from the genetic analysis because FLG loss-of-function mutations remain ill-defined in these ethnic groups.

To assess whether the lack of association of FLG null mutations with CAD is a true negative result or a reflection of the lack of statistical power, we performed a power calculation using Quanto 1.2 (http://biostats.usc.edu/software). The CAD prevalence is 1:2000 within the Tayside area of Scotland (Dawe, 2005) and FLG-null allele frequency is 0.06 (Table 2), therefore 48 cases and 100 controls provide >75% power to detect an OR of 3.2 or greater.

Chronic actinic dermatitis generally occurs on the background of pre-existing dermatitis, but the mechanisms triggering photosensitivity remain unclear. The chromophore(s) and the nature of UV-induced neoantigen initiation of a delayed hypersensitivity reaction remain unknown (Hawk, 2004). Profilaggrin and filaggrin are multifunctional proteins contributing to epidermal barrier formation and function (Brown and McLean, 2012), limiting allergen penetration and facilitating photoprotection. There is growing evidence that FLG mutations are strongly associated with delayed-type hypersensitivity (Thyssen et al., 2013) therefore FLG represents a candidate gene in CAD. The lack of association in this study has effectively excluded FLG null genotype as playing a strong role (OR >3.2) in CAD pathogenesis, and further research is needed to define key pathomechanisms. Our data represent one of the largest collections of carefully phenotyped cases of this uncommon but highly symptomatic photodermatosis. Despite increasing clinical understanding of CAD, knowledge of mechanisms and thus of informed treatment options remains limited and further studies are indicated.

Abbreviations

CAD

Chronic actinic dermatitis

CI

Confidence interval

FLG

Gene encoding filaggrin

MED

Minimal erythema dose

OR

Odds ratio

UCA

Urocanic acid

UVA

Ultraviolet A

UVB

Ultraviolet B

UVR

Ultraviolet radiation