Most cases of EH are caused by HSV-1. Because HSV seropositivity is high in the general population (~20% of children and ~60% of the adult population) it is unlikely that EH episodes are simply a function of viral exposure.⁴ An analysis of AD cases with a history of eczema herpeticum (ADEH+) examined at a single German University between 1959 to 1986 demonstrates a significant increase in the incidence of this complication, from a rate of 0.6 cases/year to greater than 15 cases/year.⁵ This increase is not likely explained by an increased prevalence of AD, since this would predict a mere doubling or tripling of the cases. Rather, it suggests that AD has evolved into a disease with greater susceptibility to infection than was observed previously, which may be the result of changes in the environment, susceptibility genes or treatment approaches. Consequently, there is growing concern that smallpox vaccination would pose a greater problem than would be explained by the increasing AD prevalence data alone.

The demographics of the subgroups (ADEH+, ADEH- and CTL) revealed significant differences in age and gender (see Table E2 in the Online Repository). Therefore, where appropriate, we adjusted for age and gender in our analysis (e.g. EASI, RL, total IgE, total eosinophil count, serum biomarkers). All diagnostic groups had the same age restrictions (1 to 80 yrs), although the ADEH+ subjects were significantly younger than the ADEH- and CTL (p<0.001; see Table E2 in the Online Repository), is likely the consequence of two factors. The first being that EH episodes typically occur early in life and therefore it was easier to find the necessary documentation of EH if the subject had experienced this complication more recently (see Table E1 in the Online Repository). The second factor was that the ADVN Genetics study that followed the Registy study restricted the age of ADEH- and CTL groups to 18 to 80 yrs to provide greater assurance that these populations had been exposed to HSV and minimizing the possibility that the difference between ADEH+ and ADEH-subjects simply reflected viral exposure. Most ADVN sites tried to enroll subjects in both Registry and Genetic studies and that would result in older subjects in ADEH- and CTL groups (see Table E2 in the Online Repository). We believe that having recruited older ADEH- subjects was a strength in that their likelihood of being misclassified would be diminished because most episodes of EH occur within the first three decades of life.⁵ There were no restrictions on gender and the ADEH+ group was equally represented by males and females. This result is consistent with previous reports showing no gender bias in EH.³ It is unclear why ADEH- subjects were more commonly female (p<0.001; see Table E2 in the Online Repository). The differences in ethnicity and race observed are likely due to the restrictions placed on the ADEH- and CTL groups that were enacted approximately one year after commencement of Registry recruitment to ensure that these two groups would also qualify for the ADVN Genetics study. For both studies, the ADEH- and CTL subjects had to self-report as non-Hispanic and either AA or EA to meet enrollment criteria. These restrictions were put in place because the ADVN Genetics study focused its initial analysis on these ethnic and racial groups to allow for smaller sample sizes while maintaining power to detect differences between ADEH+ and ADEH- populations (manuscript submitted).

Our results show that EH recurs in about half of subjects, which is more than was noted in previous publications which reported recurrence in 13 to 16% of cases.^3,5 The mean age of our ADEH+ subjects is comparable to previous publications and therefore this is unlikely to explain this difference. About 95% of ADEH+ subjects had a positive serology for HSV-1 or -2 or both (see Table E3 in the Online Repository). The majority (91%) of ADEH+ subjects were positive for HSV-1 with only 9% positive for HSV-2, confirming that most EH episodes are caused by HSV-1. This degree of HSV-1 seroprevalence is markedly higher than the most recent US National Health and Nutrition Examination Surveys (NHANES) data where the seroprevalence in the 20 to 30 years age group is only 52%.⁴ This also suggests that EH is not likely due to a diminished immunoglobulin response to HSV and is consistent with a prospective study demonstrating similar T cell and immunoglobulin responses to a diptheria-tetanus-toxoid immunization in atopic compared to noatopic subjects,⁼⁶ In contrast, only about half of ADEH- and CTL subjects were HSV-1 positive, although 31 to 36% were HSV-2 positive, which likely reflects their older age (36.0 and 38.4 yrs, respectively). Only six ADEH+ subjects had a negative serology to both HSV-1 and -2. Whether this indicates that these subjects have been missclassified as ADEH+ or the fact that some subjects were enrolled during their first EH episode and therefore their IgG response had not yet developed is not known.

One of the more remarkable findings was that ADEH+ subjects also suffered from other cutaneous infections, such as those caused by S. aureus (78% vs. 29%, ADEH+ vs. ADEH-; p<0.001), molluscum contagiosum (8% vs. 2%, p=0.001), and history of HSV infection of the eye (16% vs. 0.8%, p<0.001)(Figure 5). However, they did not have a greater incidence of dental infections (gingivitis or periodontitis) or skin infections with human papilloma virus. The frequency of S. aureus infections in ADEH+ subjects is much higher than the 30% reported in AD and observed in our ADEH- subgroup.¹² This work suggests that some global defect in cutaneous immune responses to microbes may be present in subjects with a history of EH that is relevant for both viral and bacterial infections of the skin and possibly the eyes. Interestingly, this susceptibility to S. aureus infections was also reflected in the frequency of positive Immunocap^® tests to specific staphylococcal toxins ([SEA: 43% vs. 19%, ADEH+ vs. ADEH-; p=0.001], [SEB: 43% vs. 20%, -; p=0.001], [TSST-1: 44% vs. 21%, p=0.001]).

Another signature of the ADEH+ subgroup was the breadth and magnitude of their allergen responsiveness. AD subjects underwent eight Immunocap^® teststo animal dander, food, mite-cockroach, grass, weed, mold, tree in addition to the S. aureus toxins. When Immunocap^® results were evaluated as Gaussian curves, ADEH+ had greater reactivity to six out of the eight allergen-specific Immunocaps^® compared to ADEH- subjects (Figure 4). When Immunocap^® results were analyzed as binary traits (positive vs negative), all allergens were more frequently positive in ADEH+ subjects except grass (Figure 4). The most significant differences were observed for food and perennial allergens (animal dander and mite-cockroach) suggesting that these allergens would be most predictive of ADEH+ subjects. Importantly, the greater reactivity to food allergens observed in ADEH+ subjects corroborates self-reported histories of food allergy, which were higher in this group (Figure 3). Although this is the most extensive assessment of allergen sensitization in ADEH+ subjects, a smaller study by Peng et al., showed similar findings for five allergen-specific RASTs.¹³

The greater allergen sensitization observed in ADEH+ subjects likely reflects greater Th2 polarity. To address this we evaluated several Th2 biomarkers including total IgE, eosinophil count and TARC/CCL17 (Figure 1 and ​and66).^{14, 15} TARC is a Th2-chemokine which binds to CCR4 which is highly expressed on skin-homing lymphocytes. AD subjects express high levels of TARC in lesional skin and serum levels may reach the ng/ml range as was the case in our subjects.^{16, 17} CTACK/CCL27 plays a role in the homeostatic migration of memory T cells to the skin. But CTACK is not selective for a T cell subset as serum levels are elevated in both AD and psoriasis, although CTACK levels have only been shown to correlate with disease severity in AD as was the case in our subjects (Figure 6).^{18, 19} All three Th2 biomarkers were elevated in ADEH+ compared to ADEH- subjects (p≤0.02), firmly establishing the importance of Th2 cytokines as a risk factor for widespread HSV infections in AD subjects. Wollenberg et al, demonstrated that high IgE levels were a risk factor for EH among 45 ADEH+ cases.³ Furthermore, the strong correlation between total IgE, eosinophilia and TARC with disease severity suggests that the degree of Th2 polarization is an important predictor of AD disease activity.

We found a history of food allergy and asthma was more frequently elicited from ADEH+ (69.4% and 64.3%, respectively) than ADEH- subjects (40.1% and 44.4%, respectively; p<0.001)(Figure 3). Wollenberg et al., noted a similar trend with greater reports of asthma and hay fever in EH subjects, but these were not statistically different from their control AD population.³ The food allergy prevalence of our AD subjects (40 to 69%) was higher than previous reports that estimate IgE-mediated food allergy prevalence in children with moderate-to-severe AD to be about 30%.²⁰ It is important to note that historical accounts of food allergy significantly overestimate the true prevalence sometimes by as much as 2 to 3-fold.²² Nevertheless, we had about 75% concordance with history of food allergy and food FX5E Immunocap^® result (as a binary trait). Asthma prevalence in this ADVN group are also higher than the general U.S. population, where US prevalence estimates from 1995 were 5.7% with slightly greater values for children than adults, and higher among African Americans compared to European Americans. Asthma prevalence in children with AD is estimated to be about 25 to 30% which is less than what we observed in the ADVN AD subgroups.²¹ This difference may reflect the inaccuracies of self-reporting or may suggest that the AD subjects recruited out of tertiary referral centers may in fact have more severe disease that is more frequently complicated by reactive airways. In summary, ADEH+ are more likely to have other atopic diseases than ADEH- subjects. Additionally, both AD groups report rates of food allergy and asthma that were greater than would have been predicted from published studies. This latter point may reflect a recall bias by subjects and their caregivers. We speculate that this may reflect greater disease severity of subjects recruited from tertiary referral centers for AD. Future studies proposed as part of ADVN will need to validate these historical findings.

To evaluate whether EH susceptibility could be related to a relative reduction in Th1-associated cytokines we measured IFNβ and the interferon-induced chemokine, IP-10 in age and gender-matched serum samples (see Figure E1 in the Online Repository). IFNβ values were highly variable but no difference was observed between the AD subgroups. Similarly, there were no differences in IP-10 between the AD subgroups. Peng et al. found that IFNβ was reduced in ADEH+ compared to ADEH- subjects using a similar sample size.¹³ They did not find any differences in serum IFNα or IFNγ between AD subgroups. These findings would suggest that the T cell defect in ADEH+ subjects is primarily the enhanced expression of Th2 cytokines and not diminished Th1 cytokines. Th2 cytokines are thought to be permissive to microbial invasion on the basis of their inhibitory actions on antimicrobial proteins, epidermal barrier proteins and cell-mediated immunity.^23-31

Our study confirmed and extended the finding that EH develops in AD subjects with greater disease severity. We found that the vast majority (94%) of ADEH+ subjects developed AD before 5 yrs of age. Multiple markers of AD severity including biomarkers (total IgE, peripheral eosinophil counts, TARC and CTACK), two well-accepted clinical scoring systems (EASI and RL) and BSA affected were all significantly greater in ADEH+ subjects. Most of the biomarkers are thought to dynamically reflect disease severity with the exception of total IgE which because of its long T_1/2 is more reflective of chronic changes in disease severity. Importantly, these observations were evident even after controlling for age and gender (total IgE, eosinophil counts, TARC and CTACK). Two previous publications have noted the association with early age of onset and IgE levels.^3,13 Peng et al. demonstrated that AD subjects with a history of EH had a slightly increased severity score using the SCORAD assessment (p<0.05).¹³ In our study we found that a number of the biomarkers such as TARC, CTACK, total IgE, eosinophil count and IP-10 correlated significantly with EASI and are listed in order of the strength of this correlation (Figure 2, ​,66 and Supplemental Figure 1).

Landmark studies have demonstrated that AD subjects, particularly those with more severe disease, may have a loss of function mutations in the filaggrin gene (FLG) as has been observed in ichthyosis vulgaris (IV).³² For this reason we asked subjects or their caregivers if they had a history of any of the features found in subjects with both IV and AD (e.g. keratosis pilaris, hyperlinear palms or ichthyosis). More ADEH+ (58%) reported having one or more of these features than ADEH- subjects (42%; p<0.005). Recent studies suggest that IV, diagnosed by ichthyotic changes on the anterior tibial region, can be observed in up to 32% of AD subjects.³³ Although keratosis pilaris and hyperlinear palms are less specific for IV, they are more commonly observed in AD/IV subjects (53 and 81%, respectively) than AD subjects without IV (28 and 43%; p<0.001).³³

Finally, we measured serum total IgE and a multi-allergen ImmunoCAP^® assay called Phadiatop™ on our CTL group, to provide some measure of the allergen sensitization and Th2 polarity of this group that had no personal or family history of atopic disorders (see Table E1 in the Online Repository). Although total IgE levels were within age-specific normal values and substantially lower than the values seen in both AD subgroups (p<0.001), 48% of CTL subjects had a positive Phadiatop™. We did not perform a Phadiatop™ on AD subjects so we cannot make direct comparisons with other Registry groups. This percentage was higher than that reported in a large Italian and Swiss population where the prevalence of positive Phadiatop™ ranged from 24 to 29%, respectively.^34,35 Nevertheless, NHANES III demonstrated that more than 50% of the population has a positive skin test response to at least one allergen.³⁶ Our findings agree with previous literature suggesting that total serum IgE values are a more sensitive screening assay for atopic diseases in adults than Phadiatop™.³⁷

1

2

3

4

5