Psoriasis affects 125 million people worldwide and is a chronic inflammatory autoimmune skin disease (1) associated with high risk of several comorbidities such as metabolic syndrome and cardiovascular disease (2); comorbidities that are closely related to smoking (3). Globally, 1.14 billion people smoke and tobacco consumption is of huge concern to public health (4). Patients with psoriasis are more likely to be smokers (5) and in observational studies, smoking is associated with higher risk of developing psoriasis (6–10). Furthermore, evidence suggests that prenatal tobacco exposure is associated with psoriasis in childhood (11). Taken together, evidence suggest that smoking is a strong risk factor for psoriasis. However, whether these findings reflect an actual causal association between smoking and psoriasis or can be explained by confounding or reverse causation is unclear. The gold standard to investigate causal associations is the randomized controlled trial. However, some exposures (e.g., smoking) cannot be examined in this type of study. Mendelian randomization (MR) is an epidemiological method which serves as a good alternative (12, 13). Based on Mendel´s law of inheritance, genetic variants can be used as robust proxies for modifiable exposures (in this case smoking) and causal associations can therefore be explored in an observational setting. Normally, confounding and reverse causation are intrinsic limitations in observational studies; however, in MR studies genetic variants are used as proxies for modifiable exposures. Germline genetic variants are randomly distributed during conception and are therefore not associated with confounders. Furthermore, genes are not influenced by reverse causation because they are present from birth (12). The gene CHRNA3 is found in the nicotinic acetylcholine receptor gene cluster on chromosome 15 (14, 15). The T-allele of this genetic variant is strongly associated with nicotine dependence and cumulative smoking (15), and has therefore been used as an instrument for smoking in several previous MR studies (16–20). Using this genetic variant as a proxy for high lifelong cumulative smoking, it is achievable to elucidate the causal association between smoking and psoriasis ( Figure 1 ). In this study, we hypothesized that smoking is an independent and causal risk factor for psoriasis and tested this using observational and genetic data from more than 100,000 individuals from the adult Danish general population.

Of the 105,912 participants included, we identified 1,240 individuals with moderate to severe psoriasis of whom 592 had psoriasis before examination date and 648 developed psoriasis during follow-up. In prospective observational analyzes, median follow-up was 9 years (interquartile range 7-12). Baseline characteristics are shown in Table 1 . Compared to never smokers, former and current smokers were older, less likely to be women, and had a higher alcohol consumption. Current smokers had a lower level of education and physical activity and a higher prevalence of dyslipidemia. Former smokers had a higher prevalence of type 2 diabetes and hypertension. There were no differences in baseline characteristics according to CHRNA3 genotype after correction of multiple comparisons. We found no sign of genotyping or population sampling errors (Hardy-Weinberg equilibrium P-value = 0.84).

In this study using both observational and genetic data from more than 100,000 individuals from the adult Danish general population, we found a 2.5-fold higher risk of moderate to severe psoriasis in current smokers compared to never smokers. However, using the genetic variant CHRNA3 rs1051730 as a proxy for high lifelong cumulative smoking, we did not find indications of a causal relationship between smoking and psoriasis ( Figure 4 ).

Smoking could potentially be linked to psoriasis through several pathogenic pathways (26). Smoking causes oxidative stress which produces free radicals and induces the production of several proinflammatory cytokines involved in the pathogenesis of psoriasis (27). Additionally, smoking seems to modify the expression of several genes that confer increased risk of psoriasis such as HLA-Cw6, HLA-DQA1*0201, and CYP1A1 (26). Moreover, results from epidemiological studies suggest an association between smoking and psoriasis. In the 80s and 90s, the first studies reporting associations between smoking and psoriasis were published (28–31) and subsequently, several observational studies (especially case-control and cross-sectional studies) found that patients with psoriasis smoke more than the general population, and that smoking is associated with more severe psoriasis (5). Furthermore, prospective studies have reported an increased risk of incident psoriasis among current smokers compared to never smokers (6–8, 10) and Groot et al. reported increased risk of pediatric psoriasis in children of mothers who smoked during pregnancy, including increased risk per each additional 5 cigarettes smoked daily, indicating a dose-response relationship (11).

The above-mentioned findings from observational studies could reflect a causal relationship between smoking and psoriasis; however, the association could also be explained by residual/unmeasured confounding or reverse causation. Smoking is associated with various lifestyle and socioeconomic factors (32), including comorbid diseases (33) that are also associated with psoriasis. Although observational studies try to adjust for several potential confounders, residual and/or unmeasured confounding can never fully be ruled out, and this is also the case with reverse causation. For example, the psychological burden of having psoriasis (34, 35) may lead to increased intensity of smoking to tackle life with a chronic skin disease and thus may explain the observed dose-response relationship in cross-sectional studies. Theoretically, the gold standard of proving causality is the randomized controlled trial; however, it is not feasible nor ethical to do this with smoking and psoriasis. The MR study mimics the randomized controlled trial by using genetic variants as robust proxies for modifiable exposures ( Figure 1 ). The use of MR studies in medical science is increasing (13, 36), also in the field of psoriasis (37–42). Recently, two studies reported that smoking was a causal risk factor for psoriasis (38, 43), which we could not confirm. Wei et al (38) used summary statistics in their 2-sample MR, without access to individual-level data. Thus, smokers and never smokers could not be examined separately. As smoking can only be a cause among smokers, this might explain the different results. Zhao et al (43) used a genetic instrument, constituted of 126 genotypes, but with a F-value of 15 as opposed to 32 in this study indicating that their instrument is less strong. Furthermore, our instrument is believed to have a rather clear mode of action: the genetic variant CHRNA3 rs1051730 is found in the nicotinic acetylcholine receptor gene cluster CHRNA5-A3-B4 on chromosome 15, which codes for the cerebral nicotine receptor. This leads to a variation in nicotine craving among smokers, explaining the strong association between the genetic variant and the trait (15, 20).

Strengths of this study includes the large study population and use of the unique Danish health registries. The Copenhagen General Population Study is a prospective cohort with a sample size of more than 100,000 from the general population with extensive amount of data on each participant, including genetic data free of confounding (21, 44). The Danish National Patient Registry is one of the oldest national heath registries in the world, and includes all in- and outpatient visits at the Danish hospitals since 1977 (24). Nevertheless, some limitations should be discussed. First, we only included individuals of Danish descent to minimize risk of population stratification, possibly reducing the generalizability of our findings. Second, information on smoking was collected at the day of enrollment in the study and some participants might have quit smoking during follow-up but would be defined as current smokers in our analyses. Finally, we identified individuals with psoriasis using ICD-8 and ICD-10 codes, which mainly captures individuals with moderate to severe psoriasis (23). Individuals with mild psoriasis might therefore be in our “non-psoriatic population”, potentially diluting the observational association between smoking and psoriasis.

In conclusion, in this study using both observational and genetic data from more than 100,000 individuals from the adult Danish general population, we found that smoking is observationally associated with a 2.5-fold higher risk of moderate to severe psoriasis in prospective analyses adjusted for multiple confounding factors. However, using the genetic variant CHRNA3 rs1051730, as an instrument for high lifelong cumulative smoking, we did not find indications of a causal relationship between smoking and moderate to severe psoriasis ( Figure 4 ). These results suggest that smoking is an independent, but not a causal risk factor for psoriasis. However, even though we do not find a causal association between smoking and moderate to severe psoriasis, smoking restriction and cessation remain important to minimize risk of smoking-related comorbidities.

The datasets presented in this article are not readily available because data from the Copenhagen General Population Study are subject to protection from the national Danish Data Protection Agency and we are not allowed to share the data ourselves. However, interested researchers can contact members of the steering committee to request data access. Additional data are available upon request and requests may be made to the corresponding author. Requests to access the datasets should be directed to charlotte.sigrid.erika.naeslund.koch@regionh.dk.

The studies involving human participants were reviewed and approved by the Danish ethical committees (H-KF-01-144/01). The patients/participants provided their written informed consent to participate in this study.

All authors designed the study. CN-K and SV-K did the statistical analyses. CN-K wrote the first draft of the manuscript and created figures and tables. All authors read, revised, and accepted the final article.