A novel histopathological scoring system to distinguish urticarial vasculitis from chronic spontaneous urticaria

Abstract Background Urticarial vasculitis (UV) is defined by long‐lasting urticarial lesions combined with the histopathologic findings of leukocytoclastic vasculitis. As one of the major unmet needs in UV, diagnostic criteria are rather vague and not standardized. Moreover, there seems to be considerable overlap with chronic spontaneous urticaria (CSU), particularly for the normocomplementemic variant of UV. Therefore, this study aimed to develop a diagnostic scoring system that improves the histopathologic discrimination between UV and CSU. Methods Lesional skin sections of patients with clinical and histopathologic diagnosis of UV (n = 46) and CSU (n = 51) were analyzed (blinded to the diagnosis) for the following pre‐defined criteria: presence of leukocytoclasia, erythrocyte extravasation, fibrin deposits, endothelial cell swelling, ectatic vessels, blurred vessel borders, dermal edema, intravascular neutrophil, and eosinophil numbers and numbers of dermal neutrophils, macrophages and mast cells. Results The greatest differences between UV and CSU samples were observed for leukocytoclasia (present in 76% of UV vs. 3.9% of CSU samples; p < 0.0001), erythrocyte extravasation (present in 41.3% of UV vs. 2.0% of CSU samples; p < 0.0001), and fibrin deposits (present in 27.9% of UV vessels vs. 9.7% of CSU vessels; p < 0.0001). Based on these findings, we developed a diagnostic score, the urticarial vasculitis score (UVS), which correctly assigned 37 of 46 cases of UV and 49 of 51 cases of CSU to the previously established diagnosis. Conclusion Our results suggest that the UVS, a combined quantitative assessment of the three criteria leukocytoclasia, fibrin deposits and extravasated erythrocytes, distinguishes UV from CSU in skin histopathology. The UVS, if validated in larger patient samples, may help to improve the diagnostic approach to UV.


| INTRODUCTION
Urticarial vasculitis (UV) is a rare chronic and debilitating disease defined by long lasting urticarial lesions (>24 h) and histopathological findings of leukocytoclastic vasculitis. 1 UV skin lesions come with burning or pain rather than pruritus and often resolve with purpura or hyperpigmentation. 2 The clinical spectrum of UV shows high intraindividual and interindividual variations. [3][4][5][6] Systemic manifestations, such as joint involvement with arthralgia and joint stiffness, are common; and also pulmonary, gastrointestinal and renal involvement may occur. 7,8 Those symptoms are commonly associated with hypocomplementemic UV, a rare variant of UV, which is most often associated with anti-C1q autoantibodies. 3,9 The prevalence of hypocomplementemic UV in UV patients was reported to range between 9% and 21%. 3,10,11 Some cases are linked to immune-mediated diseases such as lupus erythematosus and Sjögren's syndrome or chronic infections (e.g., hepatitis B/C, Epstein-Barr-virus and borreliosis), but in the majority of UV patients no underlying disease is identified. 1,12 One of the major challenges in UV, especially in its normocomplementemic variant, is the difficulty to distinguish it from chronic spontaneous urticaria (CSU). 13 CSU presents characteristically with recurrent itching wheals with a duration >24 h, with a disease course longer than 6 weeks. 14 Clinically, there is considerable overlap between the two diseases. Both can come with recurrent wheals and angioedema. Wheals that last longer than a day and leave transient purpura and changes in skin pigmentation upon remission are seen as signs that point to UV. 15 Moreover, CSU also sometimes shows this phenotype, especially in cases of high disease activity. 5,16 As a result, misdiagnosis and delay in diagnosis is common in UV. 6,16,17 This results in inadequate and inefficient treatment, as the first and second line treatment of CSU, standard and higher than standard doses of antihistamines, are usually not effective in UV. 15 Therefore, early diagnosis is crucial for patients with UV, and the current guideline for managing CSU recommends, when UV is suspected, performing a skin biopsy to confirm the diagnosis. 18 The histopathological evaluation of UV relies on a constellation of features including leukocytoclasia, erythrocyte extravasation, fibrin deposits, an inflammatory infiltrate of either neutrophils or lymphocytes and endothelial cell swelling. [3][4][5][6]8,10,12,16,17,[19][20][21][22][23]  Patients fulfilled the following clinical and routine diagnostic criteria: For CSU: -Recurrent spontaneous pruritic wheals (with or without angioedema) for >6 weeks consistent with a clinical diagnosis of CSU.
-Response to approved urticaria treatment (standard-dosed or updosed antihistamines or omalizumab).
-No symptoms of associated systemic disease such as arthralgia, fever attacks, hypocomplementemia.
-Routine histology of lesional skin consistent with urticaria showing no signs of vasculitis.
For UV patients: -Recurrent spontaneous pruritic or burning wheals (with or without angioedema) for >6 weeks with longer lesional duration followed by transient purpura or hyperpigmentation.
-Insufficient response to standard-dosed or up-dosed antihistamines (persistence of moderate to severe symptoms as reported by the treating physician following at least a four weeks course of treatment).
-Routine histology of lesional skin consistent with UV.
The study was approved by the local ethics committees of the universities (EA4/005/15; EA4/108/18) and patients provided written and oral informed consent.

| Selection of pre-defined histologic criteria
Aiming at developing a diagnostic algorithm for UV, a set of predefined histopathologic criteria was created after an extensive literature review (Table 1)

| Statistical analysis
In contrast to these significant differences, other criteria did not prove as a reliable tool to distinguish UV and CSU. Endothelial cell swelling was seen in 76.1% and 43.1% of UV and CSU biopsies, respectively. On average, 10.8% (±9.7%) of vessels in UV showed endothelial cell swelling as compared to 2.9% (±4%) in CSU (p < 0.0001; Figure 2D). The prevalence of ectatic vessels was not significantly different (UV: 14.8 ± 10.7% of vessels vs. CSU: 10.7 ± 9.8%), and at least one ectatic vessel was found in 82.6% of UV and 78.4% of CSU samples ( Figure 2E). All UV and 94.1% of CSU patients had at least one vessel with blurred borders in the superficial dermis, with 58.8% (±23.4%) and 26.5% (±22.2%) of vessels affected in UV and CSU patients, respectively (p < 0.0001; Figure 2F). Superficial dermal edema was seen in 54.3% of UV patients and in 68.6% of CSU patients (p = 0.148; Figure 2G).
There were no statistically significant differences in any of the histologic parameters assessed between hypocomplementemic and normocomplementemic UV patients. However, patients with hypocomplementemia showed slightly higher UVS scores (7.9 ± 4.3 vs. 5.2 ± 3.4 points, P = 0.154).

| The validity of the UVS is supported by decision tree-based discrimination of UV and CSU
To assess the validity of the UVS, we used a decision tree-based Chi-squared automatic interaction detection (CHAID) approach ( Figure 4). Using the UVS criteria, that is leukocytoclasia, fibrin deposition, and erythrocyte extravasation (with order = hierarchy), this approach diagnosed 88.7% of cases correctly, with slightly higher and lower rates than the UVS for UV (84.8%) and CSU (92.2%), respectively.

| DISCUSSION
Until now, there is no consensus on the requirements of skin histopathology to establish a diagnosis of UV. Therefore, our study aimed at refining the histopathologic discrimination of UV from its main  The comparison of clinical parameters in patients with UV versus CSU showed no major differences, however, as described earlier, 12,25 the proportion of females was considerably higher in both groups. Of

F I G U R E 4
Decision tree CHAID: The first node, leukocytoclasia, shows-if it exists-that 94.9% of positive samples belong to the UV group. If a sample shows no leukocytoclasia, but in more than 4 HPF at least 2 vessels contain fibrin, there is a 50% probability that this biopsy belongs to a patient with UV. If a sample demonstrates no leukocytoclasia, no HPF with at least 2 fibrin containing vessels and no erythrocyte extravasation, the probability to belong to the CSU group is 92.1% note, a quarter of all CSU patients declared either wheal duration Nevertheless, minor changes in CRP have also been reported in CSU as an indication for higher severity and less response to antihistamine therapy, making the distinction between UV and CSU difficult based on this criterium. 27 The histopathologic differentiation between UV and CSU is challenging and lacks standardized assessment tools. This was underlined by the limited agreement between three dermatopathologists in assigning histopathologic findings to UV or CSU. Previous studies in UV evaluated a number of histopathologic diagnostic criteria. Several authors reported leukocytoclasia and fibrin deposition as main indicators to establish a diagnosis of UV. [1][2][3][4]6,10,12,[21][22][23] In line with these findings, these two criteria demonstrated the greatest differences between UV and CSU samples in our study. However, previous studies demonstrated high variability. Leukocytoclasia, for example, was found in 22.7-75% of UV patient samples, [3][4][5] showing that our results (mean of 76%) are at the top end of the reported spectrum. The absence of leukocytoclasia 1,2,28 in all but two of our CSU samples 6,14 matches previous findings. Interestingly, fibrin deposits were found in both, UV and CSU specimens in our study, but the number of affected vessels was significantly lower in CSU compared to UV. Former studies reported fibrin in 8.8% to 88% of UV samples, [3][4][5] whereas strong fibrin deposition was seen in 1.9% of CSU patients. 29 Erythrocyte extravasation is thought to be another common finding in UV. It was reported in 17.9-77.3% of skin specimens, [3][4][5]29  showing an inflammatory infiltrate consisting of mainly lymphocytes 5 ; we did not examine this cell type in our study as this did not seem a feasible criterion for distinguishing from CSU, in which a lymphocyte-rich infiltrate is frequently found. 2,6,33 In general, it is difficult to compare the results of our study with previous reports as inclusion criteria greatly differed (e.g., no information provided about clinical presentation such as wheal duration or signs of hyperpigmentation 14  The CHAID decision tree was earlier shown to be useful in identifying independent disease predictors and supporting medical treatment decision. 34,35 In addition, the participation of two centers-which provided similar results-enhances the validity of our findings.
In conclusion, the use of a set of predefined criteria enabled us to condense the histopathologic findings that are relevant to establish a diagnosis of UV. By quantifying the criteria leukocytoclasia, intravascular fibrin and erythrocyte extravasation we provide an easy-to use diagnostic tool, the UVS, which may facilitate the histopathologic diagnosis of UV versus CSU. In order to evaluate the practicability and validity of the UVS, it should be applied to larger patient samples including both hypocomplementemic and normocomplementemic patients of different centers.

ACKNOWLEDGMENTS
The authors thank Evelin Hagen for excellent technical assistance.