No Associations Between Genetically Predicted Chronotype, Insomnia, Daytime Sleepiness, or Physical Activity and Acne Vulgaris: A Two-Sample Mendelian Randomization Study

Introduction

Acne vulgaris is a chronic skin disease featuring noninflammatory open and/or closed comedones and lesions of an inflammatory nature (eg, papules, pustules, cysts, and nodules) found on the skin, particularly the face, neck, back, chest, and upper arms.^1,2 Acne poses no risk to physical health but can have significant psychosocial effects and impact quality of life.³ Adults and adolescents with acne often have an increased risk of depression and anxiety.^4,5 Acne may also be linked to endocrinological disorders such as polycystic ovarian syndrome, precocious puberty, or rarely, late-onset congenital adrenal hyperplasia or Cushing syndrome. The exact factors contributing to the development of acne vulgaris remain poorly understood despite its high prevalence and significant impact on quality of life.⁶ While diet has long been a focus of research, lifestyle habits such as sleep and physical activity have also gained attention in recent years as potential contributors to acne pathogenesis.⁷

Indeed, there is evidence suggesting a relationship between sleep quality and acne. Studies have shown that the severity of acne is directly linked to insomnia and poor quality of life, highlighting the importance of addressing sleep disturbances in patients with acne.^8,9 Acne is also strongly associated with sleep deprivation and depressed mood, prompting physicians to consider the psychological burden of acne when managing their patients.¹⁰ Nevertheless, it remains unclear whether sleep traits have a direct causal association with acne vulgaris. Additionally, poor sleep is often linked to depression and anxiety, which are well-documented factors that negatively impact sleep quality.^11,12 This relationship becomes even more complex when considering that acne itself can induce stress and contribute to sleep disorders.¹³ On the other hand, physical activity possesses a plethora of beneficial effects on health, both physical and mental. Exercise has been shown to enhance skin health and promote anti-aging effects,¹⁴ and it may even influence certain skin conditions, such as urticaria.¹⁵ Given these benefits, it is worth investigating whether physical activity has an effect on acne vulgaris.

Establishing a causal relationship between sleep characteristics, physical activity, and acne vulgaris based solely on observational studies is challenging due to limitations such as measurement error, inherent bias, and reverse causality. To address these issues, Mendelian randomization (MR) offers a robust methodological approach.¹⁶ Genome-wide association studies (GWASs) provided results for millions of single-nucleotide polymorphisms (SNPs) and their association with various phenotypes and diseases, revolutionizing the study of complex traits.¹⁷ MR is based on the properties of the common genetic variations for different environmental exposures and allows the exploration of possible causal associations between exposures and diseases.¹⁸ Two-sample MR combines the associations between SNPs and exposure and between SNPs and outcomes from different GWASs to estimate causality.

To better understand the role of sleep traits and physical activity in acne vulgaris, we applied MR analysis to investigate the causal associations between genetically predicted sleep traits and physical activity levels and the risk of developing acne vulgaris. These findings could provide valuable insights for improving the management and care of individuals affected by acne vulgaris.

Methods

Study Design

A two-sample MR was conducted in this study, with sleep traits (chronotype, insomnia, and daytime sleepiness) and physical activity as the exposure factors and acne vulgaris as the outcome (Figure 1). Publicly available data from GWASs were used for the analysis. Ethical approval was addressed in detail in the original studies from which the data were obtained. The MR methodology is based on three key assumptions: 1) the relevance assumption (ie, the SNPs being used as IVs for the exposure are associated with it), 2) the independence assumption (ie, no common causes between the SNPs and the outcome of interest), and 3) the no horizontal pleiotropy assumption (ie, no independent pathway between the SNPs and the outcome other than through the exposure).¹⁹

Figure 1 Schematic representation of the Mendelian randomization study.

This study adhered to the “STrengthening the Reporting of OBservational studies in Epidemiology using Mendelian Randomization” (STROBE-MR) checklist guidelines to ensure comprehensive and transparent reporting standards.

Data Source

The exposure data were sourced from publicly available GWAS summary statistics obtained through the GWAS Catalog (https://www.ebi.ac.uk/gwas/). These data included the following traits within the European population: chronotype (morning person) (GCST007565; UK Biobank; 403,195 Europeans; 11,990,941 SNPs), insomnia (GCST007988; UK Biobank; 1,331,010 Europeans), daytime sleepiness (GCST012956; UK Biobank; 452,071 Europeans) and physical activity (ebi-a-GCST90093322; 89,983 individuals of European ancestry; 8,669,219 SNPs).

The GWAS summary data of acne vulgaris was obtained from the FinnGen Project (https://www.finngen.fi/en), a biobank comprising approximately 500,000 Finnish individuals of Finnish ancestry, with accessible health and SNP data. The dataset (finn-b-L12_ACNE_VULGARIS) included 1092 acne vulgaris cases and 211,139 controls, covering 16,380,453 SNPs. This dataset represents one of the most extensive collections of genetic data on acne vulgaris currently available.

To further validate our findings, we conducted an additional analysis using the GCST90245818 dataset from the Estonia Biobank (EsB), which includes 34,422 cases and 364,991 controls.²⁰ This supplementary analysis provided robust support for the reliability and consistency of our results across different populations.

Instrumental Variable Selection

The IVs were included based on the following criteria. 1) The significant SNPs related to sleep traits were selected based on P<5×10⁻⁸.²¹ For physical activity, given the limitation of obtaining an insufficient number of SNPs when applying the conventional threshold, we adopted a less stringent threshold of P < 5×10⁻⁶. This method has been widely utilized in MR studies to balance the need for sufficient IVs with the requirement for robust statistical inference, as demonstrated in prior research.^22,23 2) The SNPs were retained if they displayed a minor allele frequency (MAF) >0.01. 3) The SNPs were pruned based on the presence of linkage disequilibrium (LD) between SNPs according to R²<0.001 and a window size of 10,000 kb. 4) When an IV was absent in the summary data for the outcome, a SNP with a high LD (R²>0.8) was sought as a proxy and used for replacement. The F-value for each SNP within the IV was computed to assess the IV strength, thereby mitigating potential weak instrument bias between the IV and the exposure factor. The formula for F is F=R²×(N-2)/(1-R²). The IVs were retained if they exhibited no weak instrument bias, as indicated by F-values greater than 10.²⁴

Mendelian Randomization Analysis

The primary MR method was the inverse variance weighted (IVW) analysis. The results were presented as odds ratios (OR) and 95% confidence intervals (CIs).²⁵ The robustness of the IVW analysis was tested using the MR-Egger, weighted median, and weighted mode methods. All analyses were conducted using the “TwoSampleMR” package (version R 4.3.3). The results were visualized using scatter and sensitivity plots.

Sensitivity Analysis

Heterogeneity among IVs can affect the results of MR studies. Heterogeneity was detected using Cochran’s Q-test, where P>0.05 indicated low heterogeneity, meaning that the estimates among IVs were randomly distributed and had little impact on the IVW results.²⁶ Funnel plots were drawn to illustrate heterogeneity. The third MR assumption requires the absence of horizontal pleiotropy. The MR-Egger regression method was applied to detect the presence of horizontal pleiotropy. An intercept of the MR-Egger regression approaching zero or P>0.05 suggested the absence of pleiotropy.²⁶ The MR pleiotropy residual sum and outlier (MR-PRESSO) method was used to identify outliers (based on P<0.05) and re-estimate causal associations after their removal to correct for horizontal pleiotropy A leave-one-out analysis was performed to assess robustness and consistency by sequentially excluding each SNP.²⁷

Results

Instrument Variable Selection

In this study, 122, 33, 13, and 48 SNPs were selected for chronotype, daytime sleepiness, insomnia, and physical activity, respectively. The mean (min-max) F-values were 45.67 (29.02–168.52), 42.45 (29.51–118.15), 41.19 (30.36–94.70), and 25.09 (20.90–45.75), respectively. All F-values were >10, indicating the absence of weak instrumental bias. Thirteen SNPs were not found in the acne vulgaris dataset; proxy SNPs could not be found for two SNPs, while the others had proxies: rs9365769 was replaced by rs9356270, rs78095690 by rs6043841, rs4241964 by rs1283044, rs34719019 by rs12356674, rs2668658 by rs2732630, rs186056728 by rs117833456, rs1668835 by rs1786150, rs13059636 by rs13062430, rs12460611 by rs17599450, rs12055234 by rs60271, and rs11078398 by rs11868035. The SNPs are presented in Tables S1-S8.

Mendelian Randomization Analysis Results

The IVW method in the main analysis showed that genetically predicted chronotype (OR=1.021, 95% CI: 0.786–1.326, P=0.875), insomnia (OR=1.475, 95% CI: 0.676–3.216, P=0.329), daytime sleepiness (OR=0.466, 95% CI: 0.046–4.708, P=0.518), or physical activity (OR=0.990, 95% CI: 0.925–1.059, P=0.767) were not causally associated with acne vulgaris in the FinnGen database (Table 1, Figures 2, and S1). The results from three additional analytical methods were consistent with those of the IVW approach (Table S9).

Table 1 MR Analysis of Sleep Trait and Acne Vulgaris (IVW Results Only)

Figure 2 Scatter plots for the Mendelian randomization of acne vulgaris as the outcome and (A) chronicity, (B) daytime sleepiness, (C) insomnia, and (D) physical activity as the exposures.

To ensure the robustness of our results, we conducted a validation analysis using data from the EsB study. The IVW method in the validation analysis showed that genetically predicted chronotype (OR=0.993, 95% CI: 0.937–1.051, P=0.796), insomnia (OR=1.016, 95% CI: 0.869–1.187, P=0.843), daytime sleepiness (OR=1.045, 95% CI: 0.573–1.903, P=0.887), or physical activity (OR=0.994, 95% CI: 0.974–1.015, P=0.575) were not causally associated with acne vulgaris in the EsB Database (Table 1), supported by the other MR analyses (Table S10). These findings provide comprehensive evidence supporting the reliability of our initial screening results and strengthen the validity of the conclusions drawn from this study.

Sensitivity Analyses

Heterogeneity can affect the MR results, but Cochran’s Q-test detected no heterogeneity (all P>0.05) (Table 2 and Figure S2). No horizontal pleiotropy was detected (all P>0.05), indicating that the selected IVs met the third MR assumption (Table 2). The MR-PRESSO analysis revealed no outliers (Tables S11 and S12). The leave-one-out analysis did not show a single SNP driving the results (Figure S3). Those results are also supported by the validation analysis from the EsB (Tables 2 and S6), except for physical activity, for which heterogeneity and one outlier were detected. After removing the outlier, the results remained similar (Tables S13–S15).

Table 2 Heterogeneity and Pleiotropy Analyses

Discussion

Previous MR studies (searched on PubMed on February 11, 2025) examined acne vulgaris as an outcome of various exposures such as gut microbiota,^28–34 serum metabolites,^31,35,36 proteomics,^37,38 lipid-lowering drugs,³⁹ circulating cytokines,^40,41 lactase persistence and milk intake,⁴² and mineral deficiencies,⁴³ but none used sleep traits and physical activity as exposures. Enriching the literature using different exposures is important to better understand the pathogenesis and pathophysiology of acne vulgaris. Epidemiological studies support links between sleep traits^8–10 and physical activity^14,15 and acne (or common skin diseases). However, these studies struggled to establish causal relationships due to potential confounding factors and reverse causality. By applying MR, our study reveals no causal links between genetically predicted chronotype, insomnia, daytime sleepiness, physical activity, and acne vulgaris. Supported by sensitivity analyses and validation, these findings suggest that observed correlations may result from confounding factors or indirect pathways rather than direct causation. Clinically, this provides important insights into the etiology of acne vulgaris, helping to refine preventive strategies and focus future research on more relevant causal factors.

Several studies reported associations between sleep characteristics and acne. Still, cross-sectional and observational studies cannot determine causality. Guler et al⁹ reported that acne severity is directly correlated with insomnia and poor quality of life. Indeed, acne vulgaris is well-known to lead to image and self-esteem issues that can lead to depression and anxiety,^44,45 which in turn can lead to sleep disorders.^46,47 Sleep disorders can also lead to depression and anxiety in a vicious circle.^46,47 Still, the present study showed that the chronotype, insomnia, and daytime sleepiness were not associated with acne vulgaris. The relationship between sleep and acne may be indirect and mediated by negative emotions, but well-designed studies are necessary to examine that hypothesis.

The benefits of physical activity on physical and mental health are well known, including sleep.⁴⁸ Two previous studies reported no associations between physical activity and acne.^49,50 On the other hand, sweating can exacerbate acne due to bacteria and debris being entrapped in the pores that enlarge during physical activity and sweating.⁵¹ Still, the available evidence between physical activity and acne is scarce. Of note, physical activity is behavior influenced by one’s occupation, available time left after accounting for work and familial responsibilities, physical fitness, and willingness to exercise. Therefore, several factors besides genetics influence physical activity, while MR studies only consider the genetic component. It is a well-known limitation of the MR methodology, especially when the genetic predictability is weaker than the other factors. In addition, several environmental factors can influence the acne vulgaris risk and could not be considered because they were not included in the available datasets. Furthermore, several drugs are associated with acne and acneiform lesions, including anabolic steroids, bromides, corticosteroids, isoniazid, phenytoin, azathioprine, cyclosporine, disulfiram, phenobarbital, quinidine, tetracycline, vitamins B1, B2, B6, B12, and D2, progestin-only containing contraceptives, lithium, epidermal growth factor inhibitors (eg, erlotinib and gefitinib), and Janus kinase inhibitors (eg, abrocitinib, baricitinib, deucravacitinib, deuruxolitinib, and upadacitinib).⁵²

MR studies mimic the design of randomized controlled trials and use genetic variants as naturally occurring IVs to provide strong evidence that sleep characteristics and physical activity are not causally associated with acne vulgaris. Subsequent clinical studies need to pay attention to this point, not only focusing on sleep itself but also focusing on the impact of potential factors leading to sleep disorders on acne vulgaris, such as life stress and dietary habits.

The major advantage of this study lies in its MR design, which effectively minimizes reverse causality and confounding effects. However, several limitations warrant consideration. First, the available GWAS datasets are predominantly European, raising uncertainty about the generalizability of the findings to other populations, such as Asians. Second, GWAS participants are typically adults, whereas acne is more prevalent in children and adolescents, potentially introducing age-related biases into the analysis. Third, while MR reduces confounding, residual biases from unmeasured factors such as environmental exposures or medication use cannot be entirely ruled out. Future studies incorporating such data would enhance the robustness of causal inference. In addition, this study focused on common forms of acne vulgaris as defined by the FinnGen database and did not differentiate between subtypes due to data limitations. Future research should aim to analyze various acne subtypes, including rare forms like acne tropica and acne conglobata, to provide a more comprehensive understanding of the disease. Longitudinal datasets tracking acne severity over time would also offer valuable insights.

Conclusion

In summary, our MR study found no significant causal associations between sleep traits (chronotype, insomnia, daytime sleepiness), physical activity, and acne vulgaris. Our findings clarify previously observed associations suggesting these factors may not directly influence acne development. Future studies should explore additional genetic markers, diverse populations, and longitudinal data to validate findings and identify other causal factors for acne vulgaris.