To the Editor,
Benzoyl peroxide (BPO) is a widely used over-the-counter topic agent for treating acne vulgaris; however, the recent findings of benzene, a known carcinogen, in some BPO products have raised concerns about its safety.1,2 This study aims to assess the potential risk of skin cancer associated with BPO use in humans through a retrospective analysis.
Using data from the TriNetX US Collaborative Network, a network of 92 participating health organizations, we identified BPO exposure in patients who received a diagnosis of acne vulgaris between the ages of 12 and 70 years old and received at least 3 prescriptions containing either 2.5% or 5% BPO of any formulation. The 2.5% and 5% strengths, rather than the 10% strength, were selected to increase the likelihood of facial application of benzoyl peroxide in the exposed group. Patients who developed the outcomes of interest before the index date or within 5 years after the index date were excluded to minimize the inclusion of those with preexisting disease. This included patients with any pre-existing diagnosis code for skin cancer, benign tumors of skin, melanoma, and lipoma of the face or neck. A control comparator group was defined as patients diagnosed with acne vulgaris who had no record of receiving a prescription for benzoyl peroxide within their documented medical history between 2003 and 2024. To strengthen our results, we conducted two additional analyses using control groups of patients with seborrheic keratosis and seborrheic dermatitis, conditions associated with regular skin examinations but less commonly treated with BPO.
Outcomes of interest were limited to the face: the most common site of BPO application in the conditions studied. These outcomes included any cancer that affected the scalp, neck, ear, eyelid, or lip. The risks of various skin cancers were assessed including the risks of basal cell carcinoma, squamous cell carcinoma, and unspecified cancers of the face. In addition to cancerous outcomes, benign tumors of the face, melanocytic nevi, and lipoma were evaluated.
Propensity Score Matching (PSM) was performed for each of the three groups of exposed and control cohorts to minimize confounding and ensure comparability.3 Matching was conducted using a 1:1 greedy nearest neighbor algorithm, based on demographic characteristics such as age, sex, race, and ethnicity. Further adjustments were made using ICD-10 codes, including nicotine dependence, alcohol use, and a family history of malignancy. Additionally, the cohorts were balanced based on usage of acne treatments such as isotretinoin, tretinoin, adapalene, and tazarotene. This adjustment accounts for the photosensitizing nature of these medications, which may have effects when combined with BPO.
TriNetX's "compare outcomes" feature was used to analyze the data. The index date for the exposed cohort was defined as the point when three prescriptions for BPO and a diagnosis of acne vulgaris were documented in a patient's record. The index date for the control cohort was set to when the diagnosis of acne vulgaris first appeared in a chart. The time window for analysis began 5 years after the index date and extended over a 15-year period. Cox proportional hazards models were applied to calculate hazard ratios (HRs) and 95% confidence intervals (CIs). Censoring was applied to patients who were lost to follow-up or did not experience the outcomes of interest by the end of the study.
There was no significant difference observed in the rate of skin cancers among patients prescribed BPO compared to those without a BPO prescription, with an HR of 0.15 (95% CI: 0.04,0.56). A statistically significant decrease in the risk of benign tumors
Benzoyl peroxide (BPO) is a widely used over-the-counter topic agent for treating acne vulgaris; however, the recent findings of benzene, a known carcinogen, in some BPO products have raised concerns about its safety.1,2 This study aims to assess the potential risk of skin cancer associated with BPO use in humans through a retrospective analysis.
Using data from the TriNetX US Collaborative Network, a network of 92 participating health organizations, we identified BPO exposure in patients who received a diagnosis of acne vulgaris between the ages of 12 and 70 years old and received at least 3 prescriptions containing either 2.5% or 5% BPO of any formulation. The 2.5% and 5% strengths, rather than the 10% strength, were selected to increase the likelihood of facial application of benzoyl peroxide in the exposed group. Patients who developed the outcomes of interest before the index date or within 5 years after the index date were excluded to minimize the inclusion of those with preexisting disease. This included patients with any pre-existing diagnosis code for skin cancer, benign tumors of skin, melanoma, and lipoma of the face or neck. A control comparator group was defined as patients diagnosed with acne vulgaris who had no record of receiving a prescription for benzoyl peroxide within their documented medical history between 2003 and 2024. To strengthen our results, we conducted two additional analyses using control groups of patients with seborrheic keratosis and seborrheic dermatitis, conditions associated with regular skin examinations but less commonly treated with BPO.
Outcomes of interest were limited to the face: the most common site of BPO application in the conditions studied. These outcomes included any cancer that affected the scalp, neck, ear, eyelid, or lip. The risks of various skin cancers were assessed including the risks of basal cell carcinoma, squamous cell carcinoma, and unspecified cancers of the face. In addition to cancerous outcomes, benign tumors of the face, melanocytic nevi, and lipoma were evaluated.
Propensity Score Matching (PSM) was performed for each of the three groups of exposed and control cohorts to minimize confounding and ensure comparability.3 Matching was conducted using a 1:1 greedy nearest neighbor algorithm, based on demographic characteristics such as age, sex, race, and ethnicity. Further adjustments were made using ICD-10 codes, including nicotine dependence, alcohol use, and a family history of malignancy. Additionally, the cohorts were balanced based on usage of acne treatments such as isotretinoin, tretinoin, adapalene, and tazarotene. This adjustment accounts for the photosensitizing nature of these medications, which may have effects when combined with BPO.
TriNetX's "compare outcomes" feature was used to analyze the data. The index date for the exposed cohort was defined as the point when three prescriptions for BPO and a diagnosis of acne vulgaris were documented in a patient's record. The index date for the control cohort was set to when the diagnosis of acne vulgaris first appeared in a chart. The time window for analysis began 5 years after the index date and extended over a 15-year period. Cox proportional hazards models were applied to calculate hazard ratios (HRs) and 95% confidence intervals (CIs). Censoring was applied to patients who were lost to follow-up or did not experience the outcomes of interest by the end of the study.
There was no significant difference observed in the rate of skin cancers among patients prescribed BPO compared to those without a BPO prescription, with an HR of 0.15 (95% CI: 0.04,0.56). A statistically significant decrease in the risk of benign tumors
