Patients with atopic dermatitis (AD) harbor Staphylococcus aureus in lesional skin, causing dysbiosis, linked to disease flares and progression.¹ The onset of dysbiosis-related AD during infancy remains unclear. To clarify this, we performed a longitudinal microbiome analysis in an interventional cohort study. In total, 177 infants, all of whom received skin care, were included in this study (Figure S1A,B and Table S1). We considered the total amount of moisturizer (TAM) used as an indicator of skincare diligence. At 1 year of age, 13 infants (7.3%) developed AD. Three infants with food allergy (FA) (1.7%) were diagnosed with egg white (EW) allergy, and one of these was also diagnosed with AD. Additionally, 61 infants (34.5%) who neither had AD nor FA were sensitized to EW (EW-AS) (Figure S1B and Table S2). Compared with the healthy group, the EW-AS group received a significantly higher TAM and serum total IgE levels at 1 year of age. In contrast, the infants in the AD group received a relatively lower TAM than the healthy infant group (Table S2). The prevalence rates of vaginal delivery, exclusive breastfeeding, parental allergy history, FLG mutations, birth season, and transepidermal water loss (TEWL) levels were similar across groups, potentially due to skincare effects minimizing transepidermal water egress (Table S2).

Concerning skin microbiome, the Shannon diversity and b-diversity indices between healthy and AD groups were similar at all time points (Figure 1A,B and Figure S2). Different from our previous non-intervention cohort study,² S. aureus abundance at 6 months of age was significantly decreased compared to 1 month regardless of AD development, and the antibiotic use during the perinatal period or the mode of delivery did not affect the bacterial counts (Figure S3). The increased abundance of Streptococcus and Prevotella OTUs and the decreased abundance of Cutibacterium OTU (Cutibacterium acnes) at Day 3 after birth were associated with AD development at 1 year of age (Figure 1B,C). Spearman's rank coefficient analysis revealed a negative correlation between the TAM used (TAM 4M, TAM 6M) and the abundance of Streptococcus OTU in newborn skin (Figure 1C). In contrast, the abundance of C. acnes in newborn skin was positively correlated with TAM (TAM 1yr) and negatively correlated with TEWL levels at 1 year of age (Figure 1C). The correlation between C. acnes and the amount of moisturizer used suggests that this bacterium may proliferate by utilizing the essential lipids (cholesterol and ceramides) in the moisturizer as nutrients. These results also imply that skin dysbiosis in newborns increases the risk of AD by 1 year of age, which may be mitigated by skincare.

We also analyzed the microbial composition of healthy and EW-AS groups. We found that infants with EW-AS without AD at 1 year of age had significantly decreased skin microbiome Shannon diversity from 1 to 6 months of age (Figure 2A,B). In Day 3 infant skin, three OTUs (e.g., Pasteurellaceae, Neisseria, and Enhydrobacter) and four genera (e.g., Corynebacterium, Pasteurellaceae, Neiserria, and Enhydrobacter) showed relatively higher abundance in the EW-AS group compared with healthy group by LDA (Figure 2C). However, different from AD, the bacteria related to the development of EW-AS were not associated with TAM at any age (Figure 2C).

According to our previous Japanese infant cohort studies, the incidence rates of AD and FA around 1 year of age were between 15.3%–24.6% and 10.9%–12.1%, respectively, and food AS was observed in approximately 40% of the infants.^{3, 4} However, interestingly, in this skin care intervention study, the incidence rates of AD and FA were much lower, although the incidence rate of food AS was comparable.

In conclusion, we found infants developed AD at the age of 1 year exhibited skin dysbiosis as early as the third day of life. While the results of our study suggest a potential preventive effect of moisturizer against AD and FA during infancy, a significant limitation that requires further attention is that all patients received intervention in this study and there was no blinding involved. Future randomized trials are warranted to validate the efficacy of moisturizer-based interventions in AD and FA prevention.

特应性皮炎 (AD) 患者的病变皮肤中含有金黄色葡萄球菌，会导致生态失调，与疾病的发作和进展有关。¹婴儿期与生态失调相关的 AD 发病情况仍不清楚。为了澄清这一点，我们在一项介入队列研究中进行了纵向微生物组分析。本研究总共纳入了 177 名婴儿，全部接受了皮肤护理（图 S1A、B 和表 S1）。我们将保湿霜的总量（TAM）视为护肤勤奋程度的指标。1 岁时，13 名婴儿（7.3%）患上 AD。3 名患有食物过敏 (FA) 的婴儿 (1.7%) 被诊断为蛋清 (EW) 过敏，其中 1 名婴儿还被诊断为 AD。此外，61 名既没有 AD 也没有 FA 的婴儿 (34.5%) 对 EW 敏感 (EW-AS)（图 S1B 和表 S2）。与健康组相比，EW-AS组1岁时TAM和血清总IgE水平显着升高。相反，AD 组婴儿接受的 TAM 比健康婴儿组相对较低（表 S2）。阴道分娩、纯母乳喂养、父母过敏史、 FLG突变、出生季节和经表皮失水 (TEWL) 水平的患病率在各组之间相似，可能是由于皮肤护理作用最大限度地减少了经表皮失水（表 S2）。

关于皮肤微生物组，健康组和 AD 组之间的 Shannon 多样性和 b 多样性指数在所有时间点都相似（图 1A、B 和图 S2）。与我们之前的非干预队列研究不同，无论AD发展如何，6月龄时^2个 金黄色葡萄球菌丰度均较1个月时显着下降，且围产期抗生素使用或分娩方式不影响细菌数量。计数（图 S3）。出生后第 3 天链球菌和普氏菌OTU丰度增加以及皮肤杆菌OTU（痤疮皮肤杆菌）丰度减少与 1 岁时 AD 的发展相关（图 1B、C）。Spearman 的等级系数分析显示，所使用的 TAM（TAM 4M、TAM 6M）与新生儿皮肤中链球菌OTU的丰度呈负相关（图 1C）。相反，新生儿皮肤中痤疮丙酸杆菌的丰度与TAM（TAM 1yr）呈正相关，与1岁时的TEWL水平呈负相关（图1C）。痤疮丙酸杆菌与保湿霜用量之间的相关性表明，这种细菌可能通过利用保湿霜中的必需脂质（胆固醇和神经酰胺）作为营养物质来增殖。这些结果还表明，新生儿皮肤生态失调会增加 1 岁时患 AD 的风险，而这一风险可以通过护肤来缓解。

我们还分析了健康组和 EW-AS 组的微生物组成。我们发现，1 岁时患有 EW-AS 且无 AD 的婴儿在 1 至 6 个月大时皮肤微生物群香农多样性显着下降（图 2A、B）。在第3天婴儿皮肤中，LDA显示，与健康组相比，EW-AS组中的三个OTU（例如，巴氏杆菌科、奈瑟氏球菌和EnHydrobacter ）和四个属（例如，棒状杆菌、巴斯德氏菌科、奈瑟氏菌和EnHydrobacter）在EW-AS组中表现出相对较高的丰度（图 2C）。然而，与 AD 不同的是，与 EW-AS 发展相关的细菌在任何年龄都与 TAM 无关（图 2C）。

根据我们之前的日本婴儿队列研究，1岁左右AD和FA的发生率分别为15.3%~24.6%和10.9%~12.1%，其中约40%的婴儿存在食物AS。^{3, 4}然而，有趣的是，在这项皮肤护理干预研究中，AD 和 FA 的发生率要低得多，尽管食物 AS 的发生率相当。

总之，我们发现 1 岁时患 AD 的婴儿早在出生后的第三天就表现出皮肤生态失调。虽然我们的研究结果表明保湿霜对婴儿期 AD 和 FA 具有潜在的预防作用，但需要进一步关注的一个显着限制是所有患者都在本研究中接受了干预，并且不涉及致盲。未来的随机试验有必要验证基于保湿剂的干预措施在 AD 和 FA 预防中的功效。