The effect of habitual and experimental antiperspirant and deodorant product use on the armpit microbiome

Abstract

An ever expanding body of research investigates the human microbiome in general and the skin microbiome in particular. Microbiomes vary greatly from individual to individual. Understanding the factors that account for this variation, however, has proven challenging, with many studies able to account statistically for just a small proportion of the inter-individual variation in the abundance, species richness or composition of bacteria. The human armpit has long been noted to host a high biomass bacterial community, and recent studies have highlighted substantial inter-individual variation in armpit bacteria, even relative to variation among individuals for other body habitats. One obvious potential explanation for this variation has to do with the use of personal hygiene products, particularly deodorants and antiperspirants. Here we experimentally manipulate product use to examine the abundance, species richness, and composition of bacterial communities that recolonize the armpits of people with different product use habits. In doing so, we find that when deodorant and antiperspirant use were stopped, culturable bacterial density increased and approached that found on individuals who regularly do not use any product. In addition, when antiperspirants were subsequently applied, bacterial density dramatically declined. These culture-based results are in line with sequence-based comparisons of the effects of long-term product use on bacterial species richness and composition. Sequence-based analyses suggested that individuals who habitually use antiperspirant tended to have a greater richness of bacterial OTUs in their armpits than those who use deodorant. In addition, individuals who used antiperspirants or deodorants long-term, but who stopped using product for two or more days as part of this study, had armpit communities dominated by Staphylococcaceae, whereas those of individuals in our study who habitually used no products were dominated by Corynebacterium. Collectively these results suggest a strong effect of product use on the bacterial composition of armpits. Although stopping the use of deodorant and antiperspirant similarly favors presence of Staphylococcaceae over Corynebacterium, their differential modes of action exert strikingly different effects on the richness of other bacteria living in armpit communities.

Keywords: Antiperspirant; Armpit; Axillary region; Deodorant; Microbiology; Skin bacteria; Skin microbiome.

Figure 1. Mean composition and richness of bacterial OTUs for all three product user types, combined OTU data from two and five days after stopping product use.

Bacteria with greater than 10 sequence reads across all users in each category are shown. The top three bacterial OTUs are shown; a full list is available in Table S1. Antiperspirant users have much richer armpits (22% other bacteria versus 5% for deodorant users and 9% for no product users). At the L6 level of OTU assignment, the OTU for the highly abundant Staphylococcaceae was “Staphylococcaceae˙other” indicating that the genus was unassigned. We refer to this OTU for simplicity throughout the remainder of the figures and text as Staphylococcaceae, but that this does represent one group within Staphylococcaceae and does not denote all identified OTUs in this family.

Figure 2. Non-metric multidimensional scaling plot of armpit microbes based upon rarefaction using 1,000 sequence reads.

Small symbols represent individuals from each treatment group and large symbols represent group centroids ±1SE.

Figure 3. Mean abundances of Staphylococcaceae and Corynebacterium across product use groups.

Mean abundances of (A) Staphylococcaceae and (B) Corynebacterium of participants who regularly used antiperspirant, deodorant or no underarm products based upon sequence data. Underarm product use significantly affected the abundance of both Staphylococcaceae and Corynebacterium (2-way ANOVA: p < 0.0001 for both microbes). However, neither sampling period nor its interaction with product use significantly affected either microbe (2-way ANOVA: p > 0.05).

Figure 4. Relationship between the abundances of Staphylococceae and Corynebacterium across all individuals.

Spearman rank correlation: r = − 0.697, p < 0.0001.