Patch Testing Results From the Icahn School of Medicine at Mount Sinai Department of Dermatology 2017–2021

Allergic contact dermatitis (ACD), a type IV-mediated hypersensitivity response, is a common inflammatory skin condition, characterised by pruritus, erythema, vesicles and scaling, for which the gold standard diagnostic is patch testing [1]. Studies that comprehensively investigate the influence of demographic factors like age, sex, race and regionality are lacking. Isolating patient-specific factors that increase the risk for certain contact allergies can improve management given ACD requires an understanding of patient history and environmental exposures to determine the clinical relevance of patch test results [1]. We conducted a single-centre retrospective review of patients seen in the Department of Dermatology at the Icahn School of Medicine at Mount Sinai who received patch testing with the North American 80 Comprehensive (NAC-80) Series (Chemotechnique, Vellinge, Sweden) from 2017 to 2021. Personal information was de-identified and a waiver of informed consent was granted by the Institutional Review Board (STUDY-21-01879). Clinical and demographic information and patch test results were collected by chart review. Statistical analysis was conducted with RStudio 2022.07.1. For machine learning analysis (ML), top predictors of positive patch test results and reaction intensity were computed with Random Forest algorithm using R version 4.1.2 and randomForest package [2, 3]. %IncMSE was defined as a proxy for the predictive power of a given feature in the outcome of interest. Additional information about study methodology and results can be found in the repository: https://zenodo.org/records/13916912. Patch test results from 524 patients were included. Nickel sulphate hexahydrate (17.2%) was the most common positive allergen. Nickel sulphate hexahydrate and cobalt(II)chloride hexahydrate were the most common patches to co-occur. Univariate analyses showed significant differences in positive rates for nickel (p = 0.016), fragrance mix I (p = 0.024) and propolis (p = 0.032) based on age. Males had significantly increased incidence of positive reactions to carba mix (p = 0.003) and p-Phenylenediamine (p = 0.048), whereas females had increased incidence of positive reactions to nickel sulphate hexahydrate (p = 0.008) and cobalt II chloride hexahydrate (p = 0.021). Multivariate logistic regression models confirmed younger age was associated with testing positive for nickel sulphate hexahydrate (OR 0.98, 95% CI 0.96–0.99, p = 0.009) and cobalt II chloride hexahydrate (OR 0.97, 95% CI 0.96–0.99, p = 0.014; Table 1). Male sex was associated with a higher likelihood of testing positive for carba mix (OR 4.45, 95% CI 2.03–9.82, p < 0.001) and p-Phenylenediamine (OR 4.84, 95% CI 1.75–13.62, p = 0.002) and a decreased likelihood of testing positive for nickel sulphate hexahydrate (OR 0.33 95% CI 0.14–0.68, p = 0.005) and cobalt II chloride hexahydrate (OR 0.35, 95% CI 0.14–0.75, p = 0.013). No significant associations were identified when comparing racial groups, light skin versus skin of colour, residence (geographic location relative to NYC), comorbid atopic dermatitis (AD) or dupilumab use. Applying ML analysis identified associations with patch testing results that differed from the traditional statistical approaches. In this analysis, dark skin was the strongest variable contributing to model accuracy in predicting a positive patch test (12.38%IncMSE). Residence location (8.68%IncMSE) and race (7.05%IncMSE) were among the top three strongest predictive variables of a positive patch test. Residence (14.03%IncMSE), sex (10.47%IncMSE) and other allergic comorbidities (7.83%IncMSE) were the strongest factors predicting patch test result intensity (Table 1). Herein, we show that demographic factors like age and gender impact allergen positivity rate. Differences in nickel and cobalt sensitivity based on gender parallel previous studies. Univariate analyses also showed no association with allergic comorbidities/dupilumab use, which may be partly due to the differing immune polarisations in ACD with variable Th2 response [4]. However, ML analysis showed a potential association between allergic comorbidities and patch testing positivity/intensity, encouraging further studies to elucidate the mechanism by which allergic comorbidities, and associated therapies like dupilumab, may impact patch testing results, particularly in allergens that have more prominent Th2-driven response upon sensitisation. We are among the first to apply ML to patch testing analysis and provide important insight into the complementary nature of traditional and ML analyses [5]. The ML analysis, which captured overlapping yet distinct associations, showed that dark skin, residence and race impact model accuracy. The ML methodology captures non-linear relationships that may elucidate additional nuances in our large data set. Data on racial differences in patch test reactions remains limited and variable [6]. While some studies suggest different patch testing rates in African American patients the cause is unknown [7]. We have previously shown that residential population density is associated with higher rates of patch testing polysensitisation [8]. It is possible that location of residence modifies household size and structure, hygiene habits, indoor versus outdoor exposures and skin care habits, ultimately modifying environmental triggers and allergen sensitivity patterns. Early life environmental exposures also shape the immunologic development and may manifest with different allergen sensitisations later in life [9]. The impact of location of residence and race may also relate to differences in access to treatment, socio-economic and occupational differences. Limitations of this study include its retrospective nature, high rate of AD comorbidity due to clinical referral pattern and lack of occupational relevance assessment. Patch test readings were also performed by multiple physicians, inter-rater variability was not assessed, and readings were limited to 72 h. Nevertheless, this study provides a retrospective review of patch testing results from a large, urban medical center with a highly diverse patient population, providing the opportunity to holistically analyse the impact of patient demographics, including, sex, skin type and race, on positive patch test results. S.G., M.D., and E.D. performed data collection related to the study, critically analyzed and interpreted the data, drafted and reviewed the manuscript. R.H. critically analyzed and interpreted the data, drafted and reviewed the manuscript. M.D. and R.H. performed the statistical analyses. G.R., H.V., and S.O. participated in the data collection,interpretation of data and drafting and reviewing of the manuscript. E.G.Y. contributed to conceptualization of the study and editing and reviewing of the draft. B.U. led conceptualization of the study, interpretation of the data, drafting, editing, and review of the draft. E.G.-Y. has received research funds (grants paid to the institution) from Abbvie, Amgen, Asana BioSciences, AstraZeneca, Akari, AnaptysBio, Boehringer-Ingelheim, Eli Lilly, Galderma, Ichnos Sciences, Innovaderm Research, Janssen, Kiniksa, Kyowa Kirin, Leo Pharma, Novartis, Pfizer, RAPT Therapeutics, Regeneron Pharmaceuticals, Sienna Biopharma, UCB, Union Therapeutics and has been a consultant for Abbvie, Almirall, Amgen, Arena. Pharmaceuticals, Asana BioSciences, AstraZeneca, Aslan Pharmaceuticals, Bristol Myers. Squibb, Boehringer-Ingelheim, Cara Therapeutics, Castle Biosciences, DBV Technologies, DS. Biopharma, Eli Lilly, Evommune, Galderma, Genzyme Corporation, GSK, Ichnos Sciences, Incyte, Janssen, Kyowa Kirin, Leo Pharma, Novan, Novartis, RAPT Therapeutics, Sanofi, Sato Pharmaceutical, Siolta, Union Therapeutics and Ventyx. B.U. is an employee of Mount Sinai and has received research funds (grants paid to the institution) from Bristol Myers Squibb, Incyte, Rapt Therapeutics, Pfizer and Sanofi. He is also a consultant for Arcutis Biotherapeutics, Bristol Myers Squibb, Castle Biosciences, Fresenius Kabi, Galderma, Janssen, Lilly, Pfizer, Primus Pharmaceuticals, Sanofi and UCB. The other authors have no conflicts of interest. The study was conducted with The Icahn School of Medicine at Mount Sinai Institutional Review Board approval (STUDY-21-01879). The data that support the findings of this study are available from the corresponding author upon reasonable request.