Q & A: Cause and effect?: Derms question reported connection between moisturizers and skin cancer

Key Points

A An article published Aug. 14, 2008, in the online version of the Journal of Investigative Dermatology created quite a stir among the media. This paper, titled "Tumorigenic Effect of Some Commonly Used Moisturizing Creams when Applied Topically to UVB-Pretreated High-Risk Mice" (Yao-Ping Lu et al), drew some dubious conclusions regarding the relationship between moisturizers and skin cancer.

Since many dermatologists may be asked to comment on this research by their patients and the media, I thought it worthwhile to critically analyze the findings.

The authors state that their mouse model resembles sunlight-induced skin cancer in humans who receive heavy sunlight exposure early in life but develop skin cancers later in life in the absence of further heavy sunlight exposure.

Throughout the 17- week treatment period, most of the control mice developed tumors, and the differences from the controls were analyzed to determine whether the treatments influenced the tumor development.

The authors concluded that four commercially available moisturizers - Dermabase (Paddock Laborato- ries), Dermovan (Galderma), Eucerin Original Moisturizing Cream (Beiersdorf) and Vanicream (Pharmaceutical Specialties) - significantly increased the formation of tumors and the rate of increase in tumor size per mouse. Topical application of a custom-blended moisturizer was not tumorigenic.

There are several important problems with the conclusions drawn by the authors, the experimental model and the data analysis that make this paper inconclusive.

First, there is no evidence that tumors induced in genetically modified mice can be correlated to human skin cancer. At present, there is no reliable animal model of human skin cancer. Mice, in general, have a poorly developed horny layer that predisposes them to skin cancer.

Secondly, it is unknown whether 20 weeks of UVB exposure followed by 17 weeks of topical applications mimics human sun exposure patterns.

Thirdly, 30 mice were originally placed in each arm of the study, but a total of 14 mice failed to complete the study. There is no explanation as to what happened to these animals, which could alter the data outcome when so few animals were studied, and the observed differences (in number of tumors and increase in tumor size) from the untreated control groups were so small.

There are also some inconsistencies with the data. First, the authors combined the no-treatment and the water-treatment groups at the end of the study to increase the size of the control group to 57 animals. This manipulation created benefits for the untreated and water-treated groups in six out of nine variables analyzed, which made the moisturizer results less favorable.

Secondly, the study included data from two different experiments with identical study designs; however, the second experiment included more study groups. A comparison of the results shows wide variation in data. For example, in the Dermabase arm in experiment one, the total number of tumors was 12.39 with 1.53 standard error, and the tumor volume was 120 with a 33.00 standard error.

In the second experiment, the Dermabase arm total number of tumors was 7.52 with 1.18 standard error, and the tumor volume was 44.90 with a 23.90 standard error.

Large variability was reported in much of the study data, with a coefficient of variation for number of tumors per mouse in all but one group at greater than 84 percent.