Can atopic eczema and psoriasis coexist? A systematic review and meta‐analysis

3.1. Protocol

The review protocol is available on PROSPERO https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=118343.

3.2. Objectives of the review

The primary objective for the review was to investigate the prevalence, incidence or risk (relative risk or odds ratio [OR]) of eczema and psoriasis occurring in the same individual at the same time point. The secondary objective was to investigate the same at any time point. A tertiary objective was to summarise the immunologic and genetic data on coexistence.

3.3. Search strategy and reference searching

OVID MEDLINE and OVID Embase were searched from inception on 1st February 2020. A search strategy was built with an information specialist (D.G.) around the key terms ‘atopic eczema’, ‘psoriasis’ and ‘co‐existence’ using medical subject headings and free text words (Supporting Information 1). Reference lists of included studies were hand‐searched. Studies with a primary objective matching the review question were also included in prospective citation searching using Google Scholar. Authors of included studies were emailed to help identify any published or unpublished data on the review question.

3.4. Inclusion and exclusion criteria

Observational studies (cohort, case‐control, cross‐sectional and case‐series of at least 10 consecutive patients to give sufficient data) were included. Studies were required to provide incidence, prevalence or risk data on the coexistence of psoriasis and eczema. Cohort studies were included to provide data on coexistence in the same time period, consecutive onset and incidence estimates.

There were no restrictions on the age, gender, ethnicity, setting or geographical location of participants. Studies of all languages for which a translator was available were included in the review. The review investigated two separate populations: (i) the diagnosis of eczema (outcome) in a psoriasis population (exposure); (ii) the diagnosis of psoriasis (outcome) in an eczema population (exposure). The diseases of interest, eczema and psoriasis, were defined and recorded as a diagnosis made by a health professional (clinical diagnosis), self‐reported or the method of diagnosis was unspecified. The category of diagnosis made by a health professional was further divided into whether it was made by a dermatologist or nondermatologist, and if diagnostic criteria were used. The definition of psoriasis included all subtypes of psoriasis. The term atopic eczema has been used interchangeably with eczema/dermatitis and studies were excluded if they specifically focused on nonatopic eczema (e.g., irritant, allergic, venous, nummular). Conference abstracts were included if they contained sufficient information on the primary objective.

3.5. Study screening and eligibility assessment

Titles and abstracts were screened independently by two authors (A.C., U.A. or E.B.T.) using Rayyan QCRI (https://rayyan.qcri.org/). Full text eligibility was independently assessed by two authors (A.C., U.A. or EBT) and any differences were discussed with a third author (E.B.T. or S.G).

3.6. Data extraction

Two authors (A.C., U.A., E.B.T. or S.J.L.) independently completed the data extraction from included studies. A data extraction pro‐forma was piloted on five studies (Supporting Information 2). Timing of outcome was categorised as at the same time point (simultaneous disease), any time point (lifetime or study duration coexistence) or timing unclear. Where prevalence data were missing, study authors were contacted for further details.

3.7. Quality assessment

Included studies were critically appraised by two independent authors (A.C., U.A., E.B.T. or S.J.L.) using validated Joanna Briggs Institute critical appraisal quality assessment tools for each study type. ¹³ The number of appraisal questions for each study type differed and therefore a percentage of criteria indicating low risk of bias was calculated; ≥70% was required for studies to be classified as low risk of bias. An additional quality assessment question was added by the review team post‐hoc to assess whether the outcome diagnosis of eczema and/or psoriasis was made blinded to the review question. We felt this was an important potential source of information bias, because knowledge of the research question may have influenced whether a diagnosis of coexistent eczema and psoriasis was made or not.

3.8. Data analysis

All extracted data were included in a narrative synthesis. A meta‐analysis for the primary and secondary objectives was planned if there were sufficient clinically similar studies providing frequency/association data. Meta‐analysis was only possible for prevalence data because there were insufficient data on incidence and risk; pooled data have been presented as forest plots with 95% confidence intervals (CIs). A random effects model (DerSimonian and Laird) was used and heterogeneity assessed statistically using standard Chi‐squared (I ²). Heterogeneity was explored by planned subgroup analysis (setting, geographical area, method of diagnosis, age of participant, risk of bias). Stata v14.0 was used for the meta‐analysis. Assessment of publication bias using Egger's test was planned, but not possible because there were less than 10 studies in each meta‐analysis.

4.1. Primary objective

4.2. Secondary objective

The secondary objective investigated coexistence at any time point. Seventeen studies reported the prevalence of eczema in a psoriasis population at any time point. The main results for each study are provided in Table 1. The pooled prevalence of eczema in those with psoriasis at any time point was 9% (95% CI: 7, 10) (Figure 4). Ten studies reported the prevalence of psoriasis in an eczema population. The main results for each study are presented in Table 2.The pooled prevalence of psoriasis in those with eczema at any time point was 4% (95% CI: 3, 5) (Figure 4).

One study reported the odds and two studies reported the relative risk of eczema coexisting in a psoriasis population compared to the general population. Galili et al. ²⁴ found the adjusted OR of eczema in those with psoriasis to be 1.75 (95%CI 1.29, 2.37). ²⁴ Landgren et al. and Naldi et al. reported the relative risk of eczema in those with psoriasis to be 0.4 (95%CI 0.30, 0.55) ¹¹ and 5.5 (95%CI 3.0, 10.1) ³⁰ respectively. Both studies were rated as low risk of bias and the outcome diagnosis was made blinded, but included different populations, military conscripts versus children.

Two studies reported the odds and one study reported the relative risk of psoriasis coexisting in an eczema population compared to the general population. Zander et al. reported an adjusted OR: 0.61 (95% CI: 0.39, 0.94) and Kim et al. ²⁷ an OR 4.8 (n = 2042). The relative risk was reported as 1.41 (95% CI: 0.58, 3.46) by Williams et al. ³⁹ The two studies reporting a smaller association, Zander et al. and Williams et al. were both rated low risk of bias and the outcome diagnosis was made blinded.

4.3. Tertiary objective

Stepanova et al. ³⁵ reported immunological and genetic findings on the study population. Of individuals with a ‘mixed form’ of eczema and psoriasis, 42.5% (34/80) had an elevated total immunoglobulin E (IgE) of over 120 kU/L and females demonstrated homozygosity of the region between human leucocyte antigen B (HLA‐B) and DRB1: DRB4 (DR53). Welp et al. ³⁸ also reported laboratory findings. Of those with coexistent disease, 89% (16/18) of the patients showed a total IgE level of over 120 IU/ml and there was a significant increase in HLA‐Cw6 in 50% (9/18) patients, while HLA‐B7 and HLA‐B17 were only found in 11% (2/18) patients.

5.1. Main findings

This review provides some evidence for the coexistence of atopic eczema and psoriasis, and no convincing evidence for the complete mutual exclusivity of both diseases. The pooled prevalence results for both populations were 2%, indicating coexistence of disease may occur at a level equal or lower than expected. There were seven studies rated low risk of bias where the outcome diagnosis was made blinded to the research question, but the findings from these studies are not consistent. Prevalence at one time point may be lower than expected (0.17% for eczema in those with psoriasis and 1% for psoriasis in those with eczema), ^{10 , 39 , 40} but the relative risk for psoriasis occurring in those with atopic eczema was estimated to be 2.88. ³⁴ At any time point, results for the relative occurrence of atopic eczema in those with psoriasis and psoriasis in those with atopic eczema were conflicting, RR = 0.4 and RR = 5.5 ^{11 , 30} and 0.61 and 1.41, ^{39 , 40} respectively.

5.2. Existing literature

There is both immunological and genetic reasoning for why coexistence could be biologically plausible. Historically, it was proposed that eczema and psoriasis are driven by opposing T‐cell activation pathways. ⁴² However, the activation of Th2 and Th1 cells is likely to be antigen specific and not related to the immune system as a whole. ³⁴ Studies have indicated that activation of specific T‐cell pathways and cytokines may not be exclusive to each disease. Th17 and Th22 are typically associated with psoriasis, but have been found to have a comparatively smaller role in chronic eczema. ⁴³ In both diseases, they stimulate a keratinocyte response and epidermal thickening. This immunological crossover may be more common in children and certain disease phenotypes such as Asian atopic dermatitis. ⁴⁴

Genetically, there has been shown to be an overlap in the susceptibility foci, genetic pathways and genomic regulatory sites in allergic and autoimmune disease. ⁴⁵ For example, chromosomes 1q21, 17q25 and 20p loci identified in the UK genome screen for eczema coincide with genes which are associated with psoriasis. ⁴⁶ The area of the most overlap on chromosome 1q21 is the region which overlies the epidermal differentiation complex. Additionally, a study in this review, Stepanova et al. ³⁵ found evidence of homozygosity in the region between HLA‐B and HLA‐DRB1: DRB4 (DR53), this area determines HLA class III characteristics such as tumour necrosis factor‐α and β ²² and certain complement factors which are involved in both diseases. ³⁵

5.3. Strengths and limitations

One of the strengths of this review is the large total sample size of over 1.4 million individuals with eczema and psoriasis. The review includes studies published in English, German and French, but most studies included populations from North America and Western Europe, potentially limiting the generalisability of the findings to multiple ethnic groups. In the critical appraisal of each study, the review team agreed that a study could be defined as low risk of bias if it scored 70% or above. This threshold is arbitrary and affects which studies were included in the sensitivity analyses.

The conclusions of the review are limited by studies identified. There were important differences between studies, for example data source and methodology, and this led to high heterogeneity in the meta‐analyses. The observed heterogeneity for the pooled prevalence of eczema in those with psoriasis were heavily influenced by Abramovits et al., ¹⁵ a small cross‐sectional study of 100 patients with psoriasis. Despite looking at subgroups of studies, it was difficult to provide useful pooled values, often with sparse data or still notably different results for studies within the same strata.

Included studies used both existing data (retrospective) from departmental and national databases, and purposefully collected prospective data mostly from hospital dermatology departments. Both types of data have advantages and disadvantages. The recording of diagnoses in retrospective sources may have considered the diagnoses to be mutually exclusive, whereas unless there was blinding of the assessment, prospective collection could have been biased by the known aims of the study: leading to under or over reporting of coexistence. We have shown that the prevalence of coexistence was reported lower in studies where the outcome diagnosis was made blinded to the research question.

There were many studies with a high risk of bias (n = 21). A key critical appraisal point, and vital for the interpretation of the study, was specifying how the diagnoses was made in a valid and reliable way. We did not make assumptions if this information was not clear. Studies which used validated diagnostic criteria, or a dermatologist's diagnosis detailing criteria if no others available, provided higher quality evidence. This was more likely to occur in studies based in hospital dermatology departments. The validity of diagnoses in national databases were challenging to critically appraise and we relied on authors specifying that the diagnostic codes had been validated.

Included studies also mostly reported prevalence data and did not compare to a control group. The absence of a baseline comparator limits the interpretation of the data because population prevalence of both eczema and psoriasis differs depending on age, geographical area and setting.

The division between eczema and psoriasis may not be absolute and there is a growing body of research into ‘mixed disease’, challenging how eczema and psoriasis are categorised into distinct entities. It may also be argued that the data to support coexistence represents misdiagnosis, especially when existing data sources are used, however this does not explain the coexistence results from prospective studies assessing for simultaneous disease.

5.4. Implications

Clinicians should be aware that there is evidence for the coexistence of atopic eczema and psoriasis and be receptive to the emergence of new and evolving skin changes; potentially re‐evaluating and adding a second diagnosis of a common disease. If coexistence is identified, and the skin involvement is not controlled, this may alter the therapeutic approach. At a simple level this may include using vitamin D analogues or tar products if there is evidence of psoriasis in a patient previously diagnosed with eczema. Identifying coexistence may help direct the choice of therapies to one's that have efficacy in both diseases, for example the use of light therapy or methotrexate, or newer Janus kinase inhibitors. When using biologic therapies, it may also be useful to consider the effect these might have on differing immunological pathways for patients with coexistent diseases.