Eosinophils in psoriasis: A systematic review and meta‐analysis introducing a study quality assessment tool for diagnostic pathology studies

The histopathologic features of psoriasis are well‐documented, but recent studies have highlighted atypical features, such as eosinophils, in clinically confirmed cases.

be challenging in cases where the classical features are not distinctive, compounded by histologic changes and infiltrates not commonly expected in the disease.Dermatopathologists can aid dermatologists in differentiating one disease entity from another due to implications in management and prognosis.This is relevant in psoriasis, given the armamentarium of targeted therapy available.This systematic review and meta-analysis aimed to consolidate the data from several studies to derive a robust estimate of the prevalence of eosinophils in psoriasis.In addition, associations between prior treatment and contact dermatitis, in the form of spongiosis, and the presence of dermal eosinophils were carried out.We also introduce and apply a new quantitative Study Quality Assessment Tool that is tailored to pathology studies.

| Data collection and analysis
A systematic review of the MEDLINE (PubMed) and Scopus databases for articles published until July 2023 was performed by the lead investigator (L.C.-C.) using the search terms "eosinophils" and "psoriasis" and "biopsy."A total of 220 articles were screened.A review of titles and abstracts excluded 193 articles because they did not fulfill the inclusion criterion of quantitative assessment of eosinophils in lesional psoriasis.Additional filters for human subjects and the English language were placed and removed 20 articles.Two studies were excluded because they focused on tumor necrosis factor-alpha inhibitor-induced psoriasis.Five studies (total N = 218) were selected.
The senior author (M.A.F.) independently performed a review of PubMed and identified the same five studies.Each included study was reviewed independently by the investigators.The process of study inclusion and exclusion is summarized in Figure A1.The Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines were used to ensure transparency and completeness.
Detection was defined as the presence or absence of eosinophils.
The quantity of eosinophils was defined as the total number of extravascular eosinophils across a biopsy section.For this meta-analysis, we assumed that eosinophil counts from one study, 1 which quantified eosinophils in five representative fields, were equivalent to the eosinophil counts obtained in a representative tissue section or biopsy section.
Quantitative data were summarized as frequencies and percentages.Meta-analysis for prevalence and odds ratio (OR) was carried out using Review Manager 5.4.1 software (London, UK).The pooled prevalence and its 95% confidence interval (CI) were calculated using inverse variance with the random effect model.The pooled OR and its 95% CI were calculated using the Mantel-Haenszel method with the random effect model.Results were expressed as the number of events (N) with 95% CI and a p value of less than 0.05 considered significant.The chisquare statistic and I 2 statistic were used to detect heterogeneity; values greater than 75% indicated considerable heterogeneity.Subgroup analyses for the psoriasis subtypes were performed.

| Study quality assessment
The studies were evaluated for quality and the risk of bias prior to inclusion and analysis.We used several tools to facilitate the assessments.
The Joanna Briggs Institute (JBI) critical appraisal checklist for analytical cross-sectional studies 2 evaluates the risk of bias and decision for inclusion (Table A1).
The quality assessment tool for the case series of the National Heart, Lung, Blood Institute (NHLBI) of the National Institute of Health 3 (Table A2) included items similar to the JBI checklist.Questions regarding intervention and follow-up were not relevant to all studies.
The Newcastle-Ottawa Quality Assessment Scale was developed to evaluate the quality of non-randomized case-control and cohort studies. 4Unlike the JBI and NHLBI checklists, the Newcastle-Ottawa includes a numerical score that directly correlates to the quality of the study and may serve as the basis for its inclusion or exclusion.This scale was adapted for a cross-sectional study 5 and modified to fit a genetic urological study. 6We expanded upon the framework of the Newcastle-Ottawa adaption for cross-sectional studies and developed a Study Quality Assessment Tool for cross-sectional pathology studies (SQAT-Path) (Tables 1 and 2).

| Description of selected studies
The five studies selected were all cross-sectional single-institution case series from the United States that entailed skin biopsy analysis and review of medical records published between 2017 and 2022.All studies investigated the presence or absence of eosinophils in skin biopsies of patients with clinically proven psoriasis.The majority of the biopsies represented psoriasis vulgaris.Nearly all patients were adults, both women and men.Two studies 1,7 specified multiple psoriasis subtypes and one study focused on inverse psoriasis. 8One study 9 focused on cases with overlapping histopathologic features of psoriasis vulgaris and eczema.Table 3 summarizes the characteristics of these studies.

| Detection of dermal eosinophils
Four studies 1,8-10 observed eosinophils in almost half of the biopsy sections they examined.One study 7 reported that eosinophils were rare.
The majority of the biopsies represented psoriasis vulgaris.Two studies 1,7 specified multiple psoriasis subtypes and one study 7 In this pooled analysis, eosinophils were reported in a single case of erythrodermic psoriasis. 1Others have also identified eosinophils in both erythrodermic psoriasis and erythrodermic atopic dermatitis, additionally documenting immunophenotypic overlap between psoriasis and atopic dermatitis. 11Others have reported the presence of eosinophils in generalized psoriasis. 12r palmoplantar psoriasis, eosinophils were not reported in the two studies that had data 1,7 ; hence, pooled prevalence could not be computed.However, others have documented overlapping features of palmoplantar psoriasis and eczematous dermatitis, including eosinophils, noting that prominent eosinophilic infiltrations were more common in hand eczema compared to palmoplantar psoriasis. 13,14A B L E 1 (Continued) Thus, the presence of eosinophils in cases of erythrodermic psoriasis or palmoplantar psoriasis might reflect histomorphologic and immunophenotypic overlap with eczematous dermatitis.

| Quantification of dermal eosinophils
Upon quantification of dermal eosinophils in all five studies, the pooled prevalence of having 1 to 5 eosinophils was 24% (95% CI, 0.14-0.38)and having more than 5 eosinophils in a biopsy section was almost equally prevalent at 26% (95% CI, 0.02-0.47).
Since the presence of eosinophils in psoriasis might be attributed to superimposed allergic contact dermatitis (ACD) or other therapy, we investigated the association between treatment and the presence of eosinophils.Four studies had sufficient data to be analyzed.
There was no significant association between prior treatment and eosinophils (OR 0.52, p = 0.60).Since the data was homogenous (χ 2 = 0.88, p > 0.10; I 2 = 0), additional subgroup analysis was not performed.Post hoc power analysis was 3.3%.This was expected since the p value (0.60) was greater than α level (0.05).However, not all studies reported complete data, the pooled OR showed that prior treatment does not affect the presence or absence of eosinophils in tissue sections.Patients without prior treatment would have the same likelihood of having or not having dermal eosinophils in their biopsy.This is supported by the pooled prevalence of 47% (95% CI, 0.22-0.73)(Figure 2), which is similar to the overall prevalence of 46%.F I G U R E 1 Forest plot of the presence of dermal eosinophils in biopsy sections of psoriasis.The pooled prevalence is 0.46 or 46%, detecting eosinophils in almost half of all specimens.
Spongiosis, as a crude surrogate for ACD, was also explored.
Three studies had sufficient data for this analysis.
There was no association between spongiosis and eosinophils (OR 1.07, p = 0.29) in a homogenous dataset (χ 2 = 0.10, p > 0.05; . A limitation of this analysis is that the pooled sample sizes were too small to provide an estimate of a true effect.Post hoc power analysis yielded a value of 13.8% and is consistent with a p value (0.29) greater than α level (0.05).Due to limited data, this analysis could not focus on cases with more eosinophils, which might be more common in ACD, 15 along with greater spongiosis.The prevalence of eosinophils in specimens with spongiosis was 55% (95% CI, 0.43-0.67),slightly higher but still not too far from the overall pooled prevalence of 46% (Figure 3).

| STUDY QUALITY ASSESSMENT
Using the JBI checklist, all studies had satisfactory results (Table A1).
Most studies acknowledged the presence of confounding variables; however, none of the studies provided details on how these could be addressed.Using the NHLBI/NCI checklist for case series, all studies received a quality rating of "good" (Table A2).Using the SQAT-Path, the studies scored between 9 and 18 (out of 27).These scores translated to a quality rating of "fair" to "good" (Table 2).Areas that contributed to lower scores were the absence of details on data analysis such as sample size computation, statistical tests used, lack of identification of confounders and actions to address confounders, which, in most studies, were not pursued.Despite these limitations, the studies employed inclusion criteria and provided data that answered the research question.

| DISCUSSION
This systematic review and meta-analysis confirms that eosinophils may be present in psoriasis, but they are a variable finding (seen less than half the time overall) and, if seen, usually not often more than 5 per biopsy section.While eosinophils in even greater numbers tend to be against psoriasis, prominent eosinophils alone should not preclude the possibility of psoriasis.
When eosinophils are present, a superimposed eczematous process can be considered, especially if spongiosis is also relatively prominent; however, factors other than anti-psoriatic treatment should be assessed.
Our meta-analysis showed that the prevalence of eosinophils in lesions that had prior treatment was almost equivalent to the overall prevalence.
Our analysis also showed only a slightly higher prevalence of eosinophils in lesions with reported spongiosis compared to the overall prevalence.
Cases with overlapping clinical, histologic, immunophenotypic, and molecular features of psoriasis and atopic dermatitis, sometimes associated with targeted biologic therapy for either disease, have been increasingly recognized and can be considered in the differential diagnosis.T A B L E A 1 Study quality assessment using the Joanna Briggs Institute (JBI) checklist. 2au et al. 1 Rosa et al. 7 Penn and Brinster 10 Noorliy et al. 9 Knabel and Mudaliar 8 T A B L E A 2 Study quality assessment using the National Heart, Lung, Blood Institute of the National Institute of Health (NIH-NHLBI) checklist for case series. 3au et al. 1 Rosa et al. 7 Penn and Brinster 10 Noorliy et al. 9 Knabel and Mudaliar 8

T A B L E 1 1 .
Study Quality Assessment Tool for Cross-Sectional Diagnostic Pathology Studies (SQAT-Path) adapted from modified Newcastle-Ottawa scale. 4Diagnostic confirmation (max. of 5) (representativeness of the sample) a. Diagnosis confirmed by the external gold standard (e.g., PCR for a pathogen, metastasis, mortality; NOT based on a heuristic assessment of clinical, histopathologic, or immunophenotypic features) b.Diagnosis confirmed by delineated histologic inclusion criteria and/or clinical inclusion criteria (if relevant) c.Diagnoses clinically confirmed (if relevant) and histologically confirmed by the consensus of multiple subspecialty board-certified pathologists or equivalent in the country of origin (i.e., vs. anatomic pathology-only, dermatologyonly, or trainees) d.Diagnoses clinically confirmed (if relevant) and histologically confirmed by a subspecialty board-certified pathologist e. Diagnoses histologically confirmed but not clinically confirmed (if relevant) by subspecialty board-certified pathologist f.No description of the sampling strategy or diagnostic confirmation 2. Sample size (max. of 2) (justification and adequacy) a. Calculations based on pre-existing data b.Calculation based on estimated effect size c.No calculation/not justified 3. Sampling method (max. of 2) a.All cases that fulfill the inclusion criteria of each hospital or laboratory reporting system b.All cases that fulfill inclusion criteria from various sources (e.g., teaching files or online material with incomplete clinical records) c.No inclusion criteria 4. Sampling pool (max. of 1) a. Multi-institutional or single institution with a sample size larger than existing studies (single or multi-institution) b.Single institution Comparability (max. of 4) 1. Identification of confounding factors (max. of 2) a. Study controlled for confounding factor(s) b.Acknowledged potential confounders c.No information provided 2. Data analysis (max. of 2) a.A priori analysis planned and conducted b.Post hoc analysis conducted c.No action Outcome (max. of 13) 1.Data collection methods (max. of 2) (data readily combined with data from prior studies for pooled analysis, i.e., an effort to minimize study heterogeneity) a. Identical to pre-existing methods or no published data at the time of study design b.Deviation from published methods, justification provided c.Deviation from published methods, no justification provided 2. Assessment method (max. of 2) a. Independent blind assessment (i.e., interobserver variation calculated) b.Group/consensus assessment c.No description 3. Assessors (max. of 3) a.Multiple board-certified specialists b.Multiple observers, including board-certified specialist c.Single board-certified specialist d.No description (Continues)focused on inverse psoriasis.A subgroup analysis was conducted to check for the pooled prevalence of eosinophils based on subtype.For tissue sections taken from patients with psoriasis vulgaris, data was close to the overall prevalence with a pooled prevalence of 0.40 or 40% (95% CI, 0.24-0.56);heterogeneity was decreased (χ 2 = 23.64,p = 0.05; I 2 = 65%).For inverse psoriasis, a subtype more likely to exhibit non-classical histologic features, two studies 1,8 had a pooled prevalence of 0.70 or 70% (95% CI, 0.31-1.10),translating to detecting eosinophils in most cases of inverse psoriasis.There was no heterogeneity (χ 2 = 0.08, p > 0.05; I 2 = 0%).For guttate psoriasis, two studies 1,7 had a pooled prevalence of detecting at least one eosinophil of 0.08 or 8% (95% CI, À0.05 to 0.21).There was no heterogeneity (χ 2 =0.59, p > 0.05; I 2 = 0%).

T A B L E 3
Studies of eosinophils in psoriasis.
The publication of five studies quantifying lesional eosinophils in psoriasis in the span of 5 years (2017-2022) represented an opportunity for systematic review and meta-analysis, which have only rarely been performed in the dermatopathology literature.Information from the analysis of pooled datasets can help reconcile conflicting results, support applicability to daily practice, and guide future studies.With the advent of evidence-based medicine, systematic reviews and metaanalyses have been regarded to represent the apex of the hierarchy of scientific evidence above that of a single randomized placebocontrolled clinical trial.However, it has been noted that systematic reviews and meta-analyses may be more appropriately regarded as a F I G U R E 2 Forest plot of the presence of dermal eosinophils in biopsy sections of treated psoriasis.The pooled prevalence is 0.47 or 47%.F I G U R E 3 Forest plot of the presence of eosinophils in psoriasis specimens with spongiosis.The pooled prevalence is 0.55 or 55%.F I G U R E A 1 PreferredReporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram.
Study quality assessment for studies of eosinophils in psoriasis using the new Study Quality Assessment Tool for Cross-Sectional Diagnostic Pathology Studies (SQAT-Path).
Note: Rubric for category/domain/criteria (rationale/rubric), quality score.Quality score: Good/low risk of bias (≥18); fair/moderate risk of bias (≥8); poor/ high risk of bias (<8).T A B L E 2 1. Was the study question or objective clearly stated?Yes Yes Yes Yes Yes 2. Was the study population clearly and fully described, including a case definition?