Predictive factors for Day 7 positive patch test readings at a secondary referral centre

Summary Background Based on studies at tertiary centres it is known that patch test reading on Day (D) 7 may show additional positive reactions. Female gender, higher age and allergen groups of topicals and corticosteroids were identified as predictive factors. Objectives The first aim was to study the value of reading patch tests on D2, D3 and D7 at a secondary referral centre. The second aim was to investigate the predictive potential of the factors sex, age, atopic dermatitis, body location, allergen group and clinical relevance for a positive reaction only on D7. Methods Retrospective data from patients tested between 2013 and 2016 were evaluated. The factors sex, age, atopic dermatitis, body location, allergen group and clinical relevance were tested by regression analysis. Results Two hundred and sixty‐three out of a total of 396 patients had a positive reaction only on D2, D3 and D7 in 14 (2.5%), 152 (27.5%) and 61 (11.0%) occasions, observed in 10 (2.5%), 108 (27.3%) and 51 (12.9%) patients, respectively. These reactions were deemed relevant in 0 (0%), 12 (2.2%) and 9 (1.6%) occasions, observed 0 (0%), 11 (2.8%) and 9 (2.3%) patients, respectively. Higher age and allergen groups of metals, fragrances and resins were predictive for late positive reactions. Conclusions D7 patch test reading should also be routinely adopted at secondary referral centres. D7 positive reactions were associated with higher age and sensitization to metals, fragrances and resins.


| INTRODUCTION
Allergic contact dermatitis (ACD) is caused by delayedtype allergy (type IV hypersensitivity) to contact allergens. The pathogenesis of ACD involves an initial sensitization phase when the patient first comes in contact with the chemical. In the elicitation phase reexposure of the same chemical to the primed T-cell milieu causes release of cytokines resulting in the clinical picture of ACD. When contact allergy is suspected patch testing should be performed. It is recommended that at least two readings of the test reactions are performed. Ideally, readings are done at Day (D) 2, D3 or D4, and around D7. 1 However, in practice many dermatologists skip the late reading on D7 because it is time-consuming both for the patient and the out-patient staff. Instead, readings are performed only on D2 and D3/D4.
The value of D7 reading is based on studies at tertiary referral centres. [2][3][4][5][6][7][8][9] Contact allergens most notably causing late reactions were corticosteroids, metals and topicals such as neomycin. [2][3][4][5][6][7][8][9] The first aim was to study the value of an additional D7 reading at a large secondary referral centre. The fact that the population in secondary referral clinics is less preselected may influence the patch test outcomes. 10 Apart from scoring the reactions on D2, D3 and D7, their clinical relevance for the patient's dermatitis was estimated in order to optimize the decision on omitting one particular reading day.
To our knowledge, two studies addressed the issue of factors that may contribute to the development of D7 positive reactions. 2,3 Female gender, 2 higher age, 2,3 and allergen groups of topicals and corticosteroids 3 were identified as factors that can predict late reactions. The second aim of this study was to investigate which factors may be associated with D7 positive patch test results. The factors sex, age, past or current atopic dermatitis, body location, allergen group and clinical relevance were tested.

| Study design and patch testing
In this retrospective study, adult patients were patch tested between March 2013 and June 2016 at the Dermatology Department of St. Antonius Hospital in Nieuwegein and Utrecht, the Netherlands.
Patients were tested with the extended European Baseline Series as proposed by the Dutch guidelines on contact dermatitis. 11 The 38 allergens of this series were purchased from Chemotechnique (Chemotechnique MB Diagnostics AB).
Patch test materials were applied on the back of the patient using Van der Bend square Chambers ® (Van der Bend) and Fixomull ® tape (Beiersdorf) on D0 (Mondays or Tuesdays) and removed on D2. From January 2016, methylisothiazolinone was added to the test series. Readings were carried out on D2 (Wednesdays or Thursdays), D3 (Thursdays or Fridays) and D7 (Mondays or Tuesdays). The readings were done according to the recommendations by the European Society of Contact Dermatitis 1 which meant that reactions classified as +, ++ or +++ were reported as positive. Doubtful (?+) and irritant reactions were considered negative. A positive reaction on D2, which became negative on D3 and D7 was defined as an 'isolated' D2 ('early') reaction. A reaction that was positive only on D3 or D7 was defined as an isolated D3 reaction or D7 ('late') reaction. A positive patch test reaction was considered clinically relevant if the presence of the current dermatitis was explained by exposure to the allergen concerned. 1 Possible outcomes were 'current relevance' and 'unknown or past relevance'. It was not possible to make the subdivision into 'past' and 'unknown' relevance, because this was not always mentioned in the patient files.

| Statistics
Standard statistics were used for the descriptive analyses. Generalized linear mixed models (GLMM) regression was used to assess risk factors for an isolated D7 positive skin reaction. GLMM allows for one patient to have multiple readings. Patients were modelled as random effect by means of a random intercept. The regression analyses is performed at two hierarchical levels: at patient level and within patients at patch test reaction level. Showing the results at patch test reaction level was the only way to overcome the fact that one patient could have more positive reactions per allergen group or over more allergen groups on one particular day or on several days.

What's already known about this topic?
� Late patch test reading on Day 7 can identify new positive patch test reactions which were negative on preceding readings. This finding is derived from tertiary referral centres. � Predicting factors for positive readings on Day 7 are female gender, higher age and allergen groups of topicals and corticosteroids.

What does this study add?
� Day 7 patch test reading should also be routinely adopted at secondary referral centres. Day 2 reading may be omitted on strict conditions. � Predicting factors for Day 7 positive reactions were higher age and allergen groups of metals, fragrances and resins.
For multivariable regression analysis we reduced the allergen groups with eight categories post-hoc into six categories: metals, fragrances, resins, topicals, corticosteroids and others.
For measure of effect, we report odds ratios (OR) and 95% confidence intervals (CI). For the computations the statistical software environment R 12 was used (version 4 with the 'lme4' package 13 ).

| Occurrence of isolated D7 reaction
Initially, 623 adult patients were patch tested. Patients who were unable to attend for one or more readings on D2, D3 and D7 were excluded from the study, leaving 396 patients for further analysis. The total number of positive reactions was 553 observed in 263 out of the 396 patients.
The highest proportions of isolated D7 reactions were observed in the allergen groups topicals, corticosteroids and resins (Table 1). For absolute numbers, the groups of allergens with the highest scores were metals, fragrances and resins. Data on reaction patterns of individual allergens are mentioned in Table S3.

| Factors that may be associated with isolated D7 reaction
A summary of patient characteristics is given in Table 2. Of note is the high percentage of females (80.3%) and the preponderance of the head (62.9%) as the main affected body location.
Univariable regression analysis showed that the factors sex, past or current atopic dermatitis, and body location did not predict isolated D7 reactions (

| Occurrence of isolated D7 reactions
The first purpose of this study was to investigate the value of performing an additional D7 patch test reading at a large secondary referral centre. To our knowledge, this is the first study on this topic. The    overall prevalence figures for positive reactions to allergens on any day mirror those in the European population. 10 The lower figures for methylisothiazolinone in our study may be explained by the fact that this agent was included during the last part of the study period.
Out of the 553 positive reactions, 61 reactions were exclusively positive on D7, observed in 51 patients. Our study corroborates findings in earlier studies in which isolated D7 positive reactions were described in 4%-30% of the patients. [2][3][4][5][6][7][8][9] Our study has demonstrated the highest proportions of isolated late reactions for topicals, corticosteroids and resins. Neomycin is the most cited contact allergen for its potential to induce late reactions, 2-7,9,14 followed by corticosteroids. 1-3,6,7,14 Metals were also involved in late reactions. [3][4][5][7][8][9]14,15 Other 'late' allergens mentioned in the literature were preservatives, 2-4,7,8 colophony, 4,8 fragrances, 3,8 p-phenylenediamine, 2,5 epoxy resin, 2,3 and acrylates. 15 Our investigation confirms the results of a recent large Dutch study on the value of late reading in which the most notable late reactors in order of proportion frequency were topicals, corticosteroids, dyes, fragrances, metals and preservatives. 3 Our study has demonstrated a low percentage of isolated positive D2 readings as opposed to isolated positive D7 readings. These findings agree with those of previous studies that described the low frequency of early reactions. 4,5 Clinical relevance of a positive reaction is important given the fact that an extra reading should be of practical advantage. Relevant late positive reactions were found in 28 out of the 34 late reactions in the study by McFarlane et al. 4 In the current study 9 out 61 positive isolated D7 reactions were deemed relevant in 9 out of the 51 patients having this late reaction. The lower number of relevant reactions in our study may be caused by the fact that our population is less selected. Relevant isolated late positive reactions in our study were particularly seen in the group of fragrances.
The assessment of clinical relevance is an important tool in the interpretation of positive patch test reactions. 1 However, this may be difficult in clinical practice. In the current study, it was based on the observation that the dermatitis disappeared after cessation of contact with the suspected allergen and/or on the patient's history. The label of ingredients of cosmetic products should be checked or spot test should be performed. 1 However, in the current study ingredients of cosmetics were not checked in all cases, and no spot tests were done. Furthermore, it was not possible to make the subdivision into 'past' and 'unknown' relevance in many cases.
Notwithstanding these drawbacks, we are convinced of the reliability of clinical relevance because it was obvious from the patient's history that the majority of these reactions were of past or unknown relevance and were not contributive to their current dermatitis.
Based on the abovementioned considerations, we advise to omit the D2 reading in a selected patient group on strict conditions. Important considerations are that the baseline series is tested only, without additional test series, in patients who are able to follow the clear instructions on how to remove the patch test.

| Factors that may be associated with isolated D7 reaction
The second aim of this study was to determine the factors that were associated with a reading exclusively positive on D7. On the one hand, it was demonstrated that the factors sex, past or current atopic dermatitis, and body location did not predict late reactions. On the other hand, significant associations were found for age and the allergen groups of metals, fragrances and resins.
Our results confirmed the lack of influence of atopic dermatitis on late positive readings reported in earlier studies. 3,16 In line with two investigations, Statistics presented: n (%).

of 7 -
our results pointed to a predictive capacity of higher age. 2,3 In elderly individuals, reduced rates of thymic lymphocyte proliferation were described, which in turn resulted in reduced numbers of lymphocytes in periferal lymphoid tissues. 17 Together with other immunological changes, this caused a diminished immune function. 17 These factors may account for a slower progress of the -5 of 7 elicitation phase of the allergic contact reaction in elderly patients. Previous studies have found female gender, 2 and the allergen groups of topicals and corticosteroids 3 to be predictors of late reactions. The late reaction on corticosteroids may be explained by their antiinflammatory effect. 3 With regard to allergen group, the current study showed similar patterns, although at a lower level. Our proportion scores were highest for topicals, corticosteroids and resins, and the absolute scores were highest for metals and fragrances. Explanations for the fact that we did not find topicals and corticosteroids as predictive factors in the regression analyses may be the fact that in our study the variables were described in terms of reaction instead of patients, and the lower absolute numbers of positive test results for topicals and corticosteroids. This in turn may be caused by a different case mix in our secondary centre population as compared with the tertiary centre population in the recent Dutch study. 3 Striking features in our investigation were the high percentage of females and the preponderance of the head as the affected body site. A considerable part of our study population was composed of females with a history of atopic dermatitis having current facial dermatitis who were tested to rule out contact allergy. Schnuch et al. have found an increased proportion of females and history of atopic dermatitis in facial dermatitis. 18 Nickel and cosmetic allergens, including fragrance, were significantly more common in women than men. 18 These features may bias the selection process, which may cause the different pattern of predictive allergen groups observed in our study as opposed to the previous study. 3 Furthermore, our different case mix may potentially account for the lack of gender and body location as predictive factors for late reactions. These aberant features may indicate the different nature of a secondary referral centre.

| Strength and limitations
A novel aspect was the fact that this study proved the value of a late reading at a secondary referral centre, with its different case mix, whereas all previous studies were done at tertiary centres.
For the first time body location was used as a predictive factor for late positive readings.
Limitations are the relatively small population size and the fact that methylisothiazolinone was tested in only a small group, which could have influenced the outcomes particularly with respect to the regression analysis. Other limitations are the retrospective nature of the study, the lack of strict criteria for defining clinical relevance, and the lack of division into past and unknown relevance.

| CONCLUSIONS
In this study, it was found that D7 scoring should be routinely adopted at secondary referral centres. It was also shown that higher age, and the allergen groups of metals, fragrances and resins could predict late reactions.