Is benzene in benzoyl peroxide a cancer risk? | Lab Muffin Beauty Science (2024)

There’s been a lot of stories about benzene contamination in personal care products over the last 4 years. This year, benzene was discovered in benzoyl peroxide, a staple acne treatment. How concerned should we be?

Petition and original paper

I’ve written about the background behind Valisure’s benzene analyses before, as well as questions about the accuracy of their data and their motives.

The latest finding of benzene in benzoyl peroxide (BPO) began with a petition and a paper in March 2024, which I analysed in more detail previously. The key points:

Elevated temperatures

Valisure reported benzene levels in BPO products, but the focus was on elevated temperatures. This is an interesting choice, since Valisure’s previous petitions included lots of room temperature values.

Heat speeds up BPO decomposition. While higher temperatures are used to approximate longer periods at room temperature in accelerated stability testing, results need to be interpreted with caution, as some changes might never occur at lower temperatures. For example, 13 minutes at 75 °C is equivalent to 33 hours in a 2.5 °C fridge for accelerated stability testing, but only one of these will give a perfectly cooked sous vide egg.

The temperatures used were presented as “reasonable”, but their citations don’t really support this. Some examples:

• Most products were incubated at 50 °C for 18 days, which is supposedly a reasonable temperature “during distribution and handling by consumers” and “an accepted incubation temperature for accelerated stability studies”. However, the sources they cite indicate that 40 °C is more common for both.

• The products with the highest benzene were then heated at 70 °C for 14 to 18 days, which is “within the reported temperature range of a hot car”. Most sources say this is 50-60 °C. Their citation reports 70 °C as the maximum temperature recorded when parked in the full sun, on a day that’s over 34 °C (93.2 °F).

Benzene risks

In their petitions, Valisure use phrases to discuss benzene which, out of context, sound scary:

“Known human carcinogen” is an IARC classification that means something is conclusively linked to cancer, but this still depends on dose. This class also includes processed meat, sunlight and drinking alcohol, for example – they’re not great for you, but small exposures aren’t really worth stressing over.

“No safe level” in toxicology means small amounts will still add to the risk, but again, dose is important. There is also “no safe level” of exposure to sunlight, since any UV photon can potentially trigger cancer.

It’s also worth noting that benzene is an air pollutant, so phrases like “there is not a safe level of benzene that can exist in any skin care product” don’t make a lot of sense, since any product open to the air will contain some benzene.

Calculation confusion

The numerical comparisons they present are also a bit strange.

They compare benzene concentrations in products versus air, rather than total doses. This isn’t a great comparison, since we breathe 15.5 cubic metres of air a day, but even the most diligent skincare user uses millions of times less BPO (15.5 cubic metres is 258,333 60 mL/2 fl oz tubes).

For cancer risk, they compare “benzene in a compact car” with the amount that would cause a 1 in 100,000 increased cancer risk after inhalation for 78 years. This essentially works out what would happen if a product was left in a maximally hot car for 18 hours, then you got in and breathed that air for 78 years. This is a very unlikely scenario, but according to their data, it still only results in a rather underwhelming 0.0127% (1 in 7900) increased cancer risk.

The highest concentration they seemed to have measured at room temperature is ~18 ppm. This is worth reducing from a regulatory standpoint, but it isn’t really worth the level of panic I’ve seen – the total amount of benzene in the product would be less than what you inhale normally each day, and most of this would dissipate into the air, unbreathed.

Valisure applied for a patent on BPO-stabilising technologies in March 2023 which included some of their stability data, and the patent referenced earlier information from 2022. Since they only publicly announced this issue in March 2024, I’m not convinced that they truly believe this requires “urgent action” – there seem to be priorities other than public interest involved.

The new JID paper

Valisure have since published a new study in the Journal of Investigative Dermatology (October 2024), which has similar issues. There are three main parts: benzene measurements at room temperature, cold and hot storage of an encapsulated product, and the effect of UV exposure.

Room temperature measurements

Benzene levels in 111 BPO products were measured after storage at room temperature. The highest level measured was 35.3 ppm, which is higher than the ~20 ppm in the “Day 0” measurements from the March 2024 results.

The highest benzene concentrations were found in cleansers from Proactiv and CVS (10.85 – 35.3 ppm). The highest concentration reported in a leave-on product was 7.74 ppm (Proactiv). Most products had less than 2 ppm.

It’s worth noting that Valisure’s measurements might not be accurate, since they’ve had trouble with invalid methodologies in the past, which resulted in much higher measurements of contaminants. For example, their GCMS method decomposed the ranitidine (Zantac) they were analysing, and their LCMS method for analysing metformin measured two substances as one.

The FDA has published a paper criticising Valisure’s methods (a rare claim to fame for a “private testing laboratory”), and an FDA inspection noted “scientific deficiencies” in their methods. The US Pharmacopoeia has also published a report that seems to be pointed at Valisure, titled “Unvalidated Methods for Medicine Quality Testing Lead to Misleading Results”.

From Valisure’s petitions, it seems like they think these criticisms are irrelevant since they aren’t subject to GMP or using USP methods. But invalid methods will give invalid results, regardless of what legal framework you’re operating under. FDA representatives also met with Valisure in June 2024 “to share information about appropriate analytical testing approaches for the work that Valisure does”, which to me sounds like they don’t think their methods are appropriate.

For the time being, we’ll assume their measurements are accurate…

Encapsulated product

Epsolay (5% encapsulated BPO) was incubated at 2 °C and 50 °C. After 2 weeks, <1 ppm benzene was measured at 2 °C, and 130 ppm was measured at 50 °C.

These results aren’t surprising to me – virtually all chemical reactions are temperature-dependent. Since we already know BPO decomposes without interacting with another ingredient (this is why organic peroxides are in a special dangerous goods class), encapsulation doesn’t necessarily help.

Interestingly, they say “anecdotal evidence suggests no substantial benzene formed in the Epsolay BPO product after 2 weeks at room temperature (data not shown)”. I’m not sure what constitutes “anecdotal evidence” of benzene – the very low concentrations they’ve reported (130 ppm is 0.013%) aren’t quantifiable visually or with smell. Benzene is colourless, and was a common lab solvent for many years – you need enough evaporated molecules to be able to smell it.

UV exposure test

I’m becoming a bit of a connoisseur of Valisure’s “reasonable” conditions, so I was excited to see how they set up a UV exposure test. Like heat, UV also breaks down BPO, so we’d expect more benzene to be produced with higher UV.

Their procedure:

• 54.5 mg Proactiv Renewing Cleanser (the median sample of the most contaminated product) was spread on a 25 cm² plastic plate (about 1/12th of face area)
• It was placed in a 528 L container (roughly the volume of a refrigerator)
• It was heated to 37 °C. After 15 minutes, a UV bulb was shone on it for 2 hours, then it was left at 37 °C for another 15 minutes
• Benzene in the chamber was measured, giving a total of 14.7 µg in the air

They say this was used “to determine the extent that benzene may form when used as directed on the face and the effect of UV exposure”. In that context, their experimental choices are really interesting…

Valisure states that the UV intensity used is “one third of peak ambient sunshine”, which leads them to conclude that “reasonable sunlight levels of UV exposure may generate substantially more benzene than elevated temperature reasonably expected in transportation”. But this conclusion isn’t justified, in my opinion.

First, we need to consider the total UV dose – the number of photons hitting the BPO. Like concentration, intensity can be a misleading point of comparison – lower intensity for a longer time will give more photons to power the reaction.

Benzoyl peroxide’s absorption peak is in the short UVB region, so UVA probably isn’t doing much. If we simply look at dose, the paper reports measuring 110-120 µW/cm² UVB reaching the plate, which converts to a total dose of 8280 J/m² over 2 hours. The same dose of UVB in sunlight would be 33.12 minimal erythemal doses for Type II skin, which would not be achieved on a horizontal surface even on a full day outside, with a peak UV index of 15.

However, we should also considerwavelength. Instead of using a sunlight simulator, Valisure used a Maineng UVA/UVB bulb, which is an LED reptile lamp (I had a similar one for my pet lizard). This has three types of LEDs which produce wavelengths clustered at 307, 370 and 400 nm, rather than a continuous spectrum like sunlight. The output is unevenly weighted towards short and very long wavelengths – the paper shows a similar bulb’s spectrum overlaid on a solar spectrum (the “Vis” cutoff should be at 400 nm).

To try to strengthen Valisure’s argument, I worked out the erythemally weighted UVB. Shorter wavelengths cause more skin reddening (erythema) and more BPO breakdown, so this should be a fairer comparison.

I manually integrated the Maineng bulb spectrum from a reptile bulb database (lovingly maintained by some dedicated reptile photobiology nerds), normalised the total to 115 µW/cm², and plugged it into the erythemal weighting equation. This gave around 30.5 J/m² of erythemally weighted UVB over 2 hours, or roughly 8.8 MEDs for Type II skin.

On average, the face receives less than half the erythemal UV compared to a horizontal surface, due to its vertical orientation. 8.8 MEDs would land on the face with a full day of sun exposure, on a day with peak UV index 9 (“very high”).

This is still pretty unrealistic, since UVB doesn’t pass through glass, and leaving cleanser on your face for a full day outdoors isn’t exactly “used as directed”. There’s also plenty of ventilation outside, which would drastically reduce any inhaled benzene (benzene absorption through skin is low compared to inhalation).

(37 °C is also not the temperature of your face, it’s core body temperature – the face is usually 32-34 °C. But this part actually seems quite reasonable, relatively speaking!)

What’s the realistic risk?

So how scary is the benzene in benzoyl peroxide products?

The 2 ppm benzene that Valisure often refer to is the FDA’s limit for oral medications. This is based on an acceptable intake of 20 µg benzene a day for 70 years, resulting in 1 additional case of cancer in 100,000 exposed persons. This is very precautionary for BPO products, since benzene does not pass through skin easily (around 1% is absorbed dermally), evaporated benzene dissipates, and BPO is not typically used for an entire lifetime.

For context, we inhale around 60 µg benzene each day in urban environments, based on 2024 EPA measurements and 15.5 m³ inhaled air per day. Other high sources of exposure include cooking on a gas stove (produces 100 µg benzene) and filling up your car (9.4 to 306 µg (average 110 µg) inhaled in 5 minutes).

Given this, let’s do some very rough estimates of the benzene intake for a couple of more “reasonable” (IMO) worst-case scenarios:

Applying a leave-on product, with sun exposure

Instead of walking around outside with cleanser, let’s use the most contaminated leave-on BPO product, which had median 7.54 ppm benzene. If we scale this from 32.70 ppm, and change the applied amount to 0.5 g, 31.4 µg benzene would be generated.

Again, UVB doesn’t pass through glass or plastic, so it’s difficult to think of any situation that would be enclosed enough that you could inhale much of the benzene, AND give you UVB exposure equivalent to a full day of high UV without shade.

I can only think of sticking your head inside a sealed lizard tank for 2 hours. Assuming your lizard tank is small enough that you’d inhale every molecule of evaporated benzene, this results in absorption of 16 µg benzene (51% of 31.4 µg: 50% through inhalation and 1% through dermal absorption) – below the FDA’s 20 µg acceptable intake.

In other words, the added risk would be less than an extra 1 in 100,000 chance of cancer, even if you performed this lizard tank ritual daily for 70 years.

Using a cleanser in a sealed shower cubicle

A realistic worst-case scenario with a high benzene BPO cleanser might be using it all over your body in a completely sealed shower.

As part of the face model experiment, Valisure also measured 4 µg of benzene evaporating from 0.63 g of the second most contaminated cleanser sample at 37 °C, over 24 hours.

If we scale this up to 18.67 g of shower gel (the amount used in SCCS safety assessments) and assume that all the benzene that evaporated in Valisure’s experiment over 24 hours would evaporate during the shower, we get 118.5 µg benzene in the air.

A standard 3 x 3 x 8 foot shower is 72 cubic feet, or 2038.81 L, giving 0.05812 µg/L in the air (there’s no ventilation in your shower for some reason – the drain is clogged, I guess).

If you’re having a long, vigorous “light intensity exercise”-style shower for 30 minutes (12.4 L of air a minute), you’d inhale 372 L of air, which would have 21.62 µg benzene. Since the cleanser doesn’t stay for long on your skin, your intake would just be the 50% through inhalation, which is 10.8 µg. Again, this is below the FDA’s acceptable intake, resulting in less than a 1 in 100,000 increased risk of cancer, even if your drain is clogged for 70 years.

Evidence of actual risk?

Since the March results were released, there’s been two analyses coauthored by Dr John Barbieri, one of the top acne dermatologists, to look for evidence of elevated benzene exposure from benzoyl peroxide.

The first analysis looked at NHANES data from 1999-2018 on blood benzene levels, comparing 14 individuals who reported current benzoyl peroxide use with 65 matched controls. There was no association between benzoyl peroxide exposure and blood benzene levels.

The second study looked at healthcare patient data to see if benzoyl peroxide use was associated with cancers linked to benzene (leukemia or lymphoma) or internal malignancies over the next 10 years. The comparison group were diagnosed with nevus/seborrheic keratosis, or viral warts. This analysis included over 60,000 patients treated with benzoyl peroxide. There wasn’t any evidence of increased lymphoma, leukemia, or internal malignancies.

While these studies do have limitations, the results are exactly what we’d expect based on Valisure’s own data – you’d get minimal extra benzene exposure from benzoyl peroxide, despite all the scary headlines.

Conclusion

Again, I’m not against reducing our benzene exposure – I think it’s definitely an issue that regulators should look into, since 1 in 100,000 risks are worth minimising on a population level.

But given how much benzene we’re exposed to from other sources, and how little is inhaled from the most contaminated products even in highly-contrived worst-case scenarios (again, assuming their measurements are accurate), the dramatic announcements really don’t seem warranted. In my opinion, they seem much less geared towards improving public health, and more towards perhaps setting the stage for lawsuits, or generating demand for new products.

References

Kucera K, Zenzola N, Hudspeth A, et al. Evaluation of Benzene Presence and Formation in Benzoyl Peroxide Drug Products. J Invest Dermatol. 2024 Sep 27:S0022-202X(24)02155-9. doi:10.1016/j.jid.2024.09.009

Yang J, Marzan TA, Ye W, Sommers CD, Rodriguez JD, Keire DA.A Cautionary Tale: Quantitative LC-HRMS Analytical Procedures for the Analysis of N-Nitrosodimethylamine in Metformin.AAPS J. 2020;22(4):89. doi:10.1208/s12248-020-00473-w

Furman JP. Valisure LLC, New Haven, CT Letter (lab inspection).December 5, 2022.

United States Pharmacopoeia. Unvalidated Methods for Medicine Quality Testing Lead to Misleading Results. August 2023.

US Food and Drug Administration (FDA), Center for Drug Evaluation and Research. Minutes of Meeting with Valisure, LLC. June 3, 2024.

McKenzie RL, Lucas RM. Reassessing Impacts of Extended Daily Exposure to Low Level Solar UV Radiation. Sci Rep. 2018;8(1):13805. doi:10.1038/s41598-018-32056-3

Wunderlich S, Griffiths T, Baines F. UVB‐emitting LEDs for reptile lighting: Identifying the risks of nonsolar UV spectra. Zoo Biol. 2024;43(1):61-74. doi:10.1002/zoo.21806

Wunderlich S. Reptile Lamp Database: Spectrum 542: BCLED1. April 7, 2020.

Vernez D, Milon A, Vuilleumier L, et al. A general model to predict individual exposure to solar UV by using ambient irradiance data. J Expo Sci Environ Epidemiol. 2015;25(1):113-118. doi:10.1038/jes.2014.6

Williams PRD, Sahmel J, Knutsen J, Spencer J, Bunge AL. Dermal absorption of benzene in occupational settings: Estimating flux and applications for risk assessment. Crit Rev Toxicol. 2011;41(2):111-142. doi:10.3109/10408444.2010.530224

United States Environmental Protection Agency. Air Data: Air Quality Data Collected at Outdoor Monitors Across the US. October 29, 2024.

Kashtan YS, Nicholson M, Finnegan C, et al.Gas and Propane Combustion from Stoves Emits Benzene and Increases Indoor Air Pollution.Environ Sci Technol.2023;57(26):9653-9663. doi:10.1021/acs.est.2c09289.

Edokpolo B, Yu QJ, Connell D.Health risk assessment of ambient air concentrations of benzene, toluene and xylene (BTX) in service station environments.Int J Environ Res Public Health.2014;11(6):6354-6374. doi:10.3390/ijerph110606354

SCCS Notes of Guidance for the testing of cosmetic ingredients and their safety evaluation – 12th revision. Published December 22, 2023.

Agency for Toxic Substances and Disease Registry. Guidance for Inhalation Exposures. Department of Health and Human Services, 2021.

Sadr N, Troger A, Chai PR, Barbieri JS. No evidence for an association between benzoyl peroxide use and increased blood benzene levels in the National Health and Nutrition Examination Survey. J Am Acad Dermatol. 2024;91(4):763-765. doi:10.1016/j.jaad.2024.06.050

Garate D, Thang CJ, Lai J, Golovko G, Wilkerson MG, Barbieri JS. Benzoyl peroxide for acne treatment is not associated with an increased risk of malignancy: A retrospective cohort study. J Am Acad Dermatol. 2024;91(5):966-968. doi:10.1016/j.jaad.2024.07.011