STRENGTHS AND LIMITATIONS OF THIS STUDY

This represents the first study employing Mendelian randomisation to ascertain the causal relationship between night shift work and prostate cancer (PCa).

Two large cohorts were established for analysis: the current work schedule cohort included 130 853 participants, and the lifetime night shift work cohort included 49 511 participants.

It also stands as one of the few investigations offering a comprehensive evaluation of the impact of night shift work on PCa, considering current work schedule, frequency of night shift work and lifetime duration of night shift work.

Several unaccounted factors may have implications for the precision of our findings, such as intensity of night shifts, cumulative night shift exposure and type of night shift rotation.

The study lacked pathological data, making it impossible to conduct subgroup analyses based on PCa subtypes, which may obscure certain true associations.

Introduction

Prostate cancer (PCa) is prevalent diseases that significantly affect the health and quality of life of middle-aged and elderly men. According to global cancer statistics in 2020, PCa was the second most common cancer and the fifth leading cause of cancer-related deaths among men worldwide. In more than half of the countries worldwide, PCa is the most frequently diagnosed cancer in male group.1 Hence, additional investigation into risk factors for PCa is imperative to attain the objective of early prevention.

In recent years, shift and night shift work have garnered recognition as significant contributors to health burdens. Data from 2021, amid the conditions imposed by the COVID-19 pandemic, indicate that 21% of workers in the 27 European Union countries engaged in night work, with a higher prevalence among men (25%) than women (17%).2 Many studies have confirmed that night shift work is closely related to the incidence of hormone-related cancers such as breast cancer, endometrial cancer and PCa.3 The disruption of circadian rhythms caused by night shift work has been linked to metabolic disorders, cancer, metabolic syndrome, peptic ulcers and other diseases, as well as disorders of sex hormones.4 5 In animal experiments, it was also found that light-induced inhibition of circadian rhythms increases carcinogenesis in rodents.6 Based on ‘limited evidence for the carcinogenicity of shift work involving night work in humans’ and ‘sufficient evidence for the carcinogenicity of light during the dark period of the day (biological night) in experimental animals’, the International Agency for Research on Cancer (IARC) working group concluded that, ‘Shift work that disrupts circadian rhythms may cause cancer’ (group 2A).7 In the realm of PCa, despite numerous cohort studies8–16 and case-control17–23 investigations, a unanimous consensus regarding the heightened risk of PCa associated with night shift work remains elusive. Given the prevalence of night shift work and its substantial public health and policy implications, further in-depth research is needed. Hence, we aimed to verify the connection between night shift work and PCa by using the UK Biobank (UKB) database and employed two-sample Mendelian randomisation (MR) to assess the causal relationship.

Subjects and methodsStudy population

The UKB database enrolled a cohort of more than 500 000 participants aged 37–73 years from the UK between 2006 and 2010 (https://www.ukbiobank.ac.uk). Comprehensive baseline data, including demographic characteristics, lifestyle factors, occupational information, biological samples and medical conditions, were systematically collected at the time of recruitment. The UKB further conducted three follow-up visits. Additionally, an online follow-up questionnaire was completed by 70 480 participants in 2015 to provide detailed information regarding lifetime employment. UKB has been previously described in detail.24 Exclusion and inclusion criteria were determined based on the baseline data. Initially, female participants from the UKB were excluded, along with individuals diagnosed with any form of cancer, and those who had undergone prostatectomy or orchiectomy (figure 1). Subsequent analyses were confined to participants who reported ‘In-paid employment or self-employment’ and completed the comprehensive lifetime occupation questionnaire.

Figure 1

Study design and workflow. GWAS, Genome-Wide Association Study; MR, Mendelian randomisation; PCa, prostate cancer; SNPs, single nucleotide polymorphisms.

All the participants provided informed consent.24

Exposure and outcome assessmentShift work

In the UKB, shift work was defined as ‘a work schedule that falls outside of the normal daytime working hours of 9 a.m. to 5 p.m., and this may involve working afternoons, evenings, or nights or rotating through these kinds of shifts’ (https://biobank.ndph.ox.ac.uk/showcase/field.cgi?id=826). Night shift work was defined as ‘a work schedule that involves working through the normal sleeping hours, for instance working through the hours from 12 a.m. to 6 a.m.’ (https://biobank.ndph.ox.ac.uk/showcase/field.cgi?id=3426).

During the initial recruitment phase, participants were prompted to respond to a touchscreen query: ‘Which of the following describes your current situation?’ If participants indicated ‘In-paid employment or self-employment’, they were subsequently queried with ‘Does your work involve shift work?’. Response options included ‘Never/rarely, Sometimes, Usually, and Always’. Those respondents selecting ‘Sometimes, Usually, or Always’ were further probed with ‘Does your work involve night shifts?’ The options for this subsequent question remained consistent. Through the amalgamation of the two inquiries, we categorised the current work schedule into the following classifications: ‘Day work’, ‘Shifts but no night shifts’, ‘Some night shifts’, and ‘Usual night shifts’.

In the questionnaires pertaining to lifetime employment, the classification of night shifts involved engaging in work for a minimum of 3 hours between midnight and 05:00 (https://biobank.ndph.ox.ac.uk/showcase/field.cgi?id=22650). Participants documented all occupations they had ever undertaken, supplying details such as the duration of employment in each role, indication of whether night shifts were a component of the job and the monthly frequency of night shifts. We exclusively considered lifetime employment information predating the baseline assessment. Using these data, we computed the cumulative duration of night shift work and the mean monthly frequency of night shifts. As suggested in previous studies,25 the number of years working night shifts was divided into three categories: none, ≤10 years and >10 years. The average frequency of night shifts was also divided into four categories: none, <3/month, 3–8/month and >8/month.

Prostate cancer

The outcome, incident PCa (field ID 40006, International Classification of Diseases (ICD)-10 code C61), was obtained from cancer register (category ID in UKB 100092) through linkage to national cancer registries. National cancer registries centralised information received from separate regional cancer centres around the UK.

Covariates

All covariates encompassed in this study were categorised into three principal domains: demographic characteristics, lifestyle factors and anthropometric parameters, and sleep-related variables. Demographic information encompassed age, ethnicity (white or other), education (university/college degree or other), the Townsend Deprivation Index and the family history of PCa. Lifestyle factors and anthropometric parameters included variables such as activity time, sedentary time, weekly work hours, smoking status (never, previous, current), alcohol consumption (never, previous, current), history of insulin, cholesterol and blood pressure medication, waist circumference, fasting glucose, triglycerides, HDL cholesterol, blood pressure, C reaction protein, testosterone and sex hormone binding globulin. Additionally, sleep-related factors comprised variables related to sleep duration and chronotype (morning, intermediate and evening). The definition of chronotype was derived from the UKB questionnaire. Participants were asked, ‘Which one of these types do you consider yourself to be?’ The response options included ‘Definitely a morning-type’, ‘Rather more a morning-type than an evening-type’, ‘Rather more an evening-type than a morning-type’, and ‘Definitely an evening-type’. For further analysis, the two intermediate categories were grouped as ‘intermediate’.

Statistical analyses

Continuous variables were expressed as means (SD), and categorical variables were presented as frequencies (percentages). Missing values for continuous variables underwent imputation through multiple imputations, while missing values for categorical variables were treated as an ‘unknown’ category. The assessment of associations was conducted using Cox proportional hazards models to calculate HR and corresponding 95% CI. Initially, we explored the relationship between night shift work and PCa in current work schedule. Then, we investigated the association between average lifetime night shift frequency and cumulative lifetime duration of night shifts with PCa. Three distinct models were employed for analysis. Model 1 was an unadjusted model. Next, we used directed acyclic graph analysis to identify potential confounders, which included age, ethnicity, education and socioeconomic status (using the Townsend Deprivation Index as a proxy) (online supplemental figure S1). Consequently, model 2 was adjusted for these identified confounders. Additionally, based on previous research,9 19 23 26 model 3 included multivariable adjustments, which encompassed age, ethnicity, education, Townsend Deprivation Index, family history of PCa, body mass index (BMI), activity time, sedentary time, weekly work hours, smoking status, alcohol consumption, sleep duration and chronotype. In the adjustments, age, ethnicity, education, family history of PCa, smoking status, alcohol consumption and chronotype were treated as categorical variables, while age, Townsend Deprivation Index, BMI, physical activity time, sedentary time, weekly work hours and sleep duration were treated as continuous variables.

Considering the substantial impact of age on PCa onset, we performed a stratified Cox regression analysis, categorising age into the following subgroups: 36–45, 46–55, 56–65 and 66–75 years. Subsequent subgroup analyses were conducted based on chronotype. In order to further address potential biases introduced by other risk factors for PCa, additional subgroup analyses were performed on BMI (<25 kg/m2, 25≤BMI <30 kg/m2 and<25 kg/m2), family history of PCa (https://www.cancerresearchuk.org/about-cancer/prostate-cancer/risks-causes).

Statistical significance was determined using a two-tailed test at p<0.05. P values for trends were computed based on the categorical variables of current night shift work, lifetime duration and frequency. Statistical analyses were performed using R software (V.4.3.2).

In the sensitivity analysis, we removed all missing values and repeated the main analysis. The control group consisted of day workers in all analyses.

Two-sample MR analysis

In the cohort study, despite implementing multifactorial regression analysis and subgroup analyses, our findings remain vulnerable to potential confounding factors. In light of this, for enhanced verification of the robustness and validity of our study outcomes, a two-sample MR analysis was conducted to investigate the causal link between night shift work and the risk of PCa. MR leverages genetic variations as instrumental variables to scrutinise whether a causal association exists between the exposure and outcome variables over time. Given the random allocation of genes at conception on a population level, MR analysis serves to minimise the impact of confounding factors to the utmost extent.27 Comprehensive details regarding the MR analysis are elucidated in the supplementary methods section of the online supplemental materials.

The patient and public involvement statement

The UKB database was used directly in this investigation. No direct involvement of patients or the public was observed in the course of this study.

ResultsCharacteristics of the study population

Over a median follow-up duration of 13.9 years, the current work schedule cohort recorded 4993 cases of PCa. Additionally, within the lifetime night shift work cohort, 2022 cases of PCa were documented over a median follow-up period of 14.0 years. In the study population focusing on PCa, the mean lifetime night shift frequency was 8.62±6.05 times, and the cumulative lifetime duration of night shifts was 20.86±13.73 years.

In comparison to individuals engaged in daytime employment, participants involved in night shift work exhibited a younger age profile, a predominantly non-white demographic and higher BMI. Those with a higher frequency of night shift work demonstrated an elevated Townsend Deprivation Index, lower educational attainment, extended working hours, prolonged sedentary periods and shorter sleep duration. Evening chronotype was more prevalent among workers engaged in night shift work. Night shift workers exhibited a higher prevalence of smoking but a lower prevalence of alcohol consumption (table 1). Baseline characteristics data of the research population, categorised based on their lifetime occupation reports, are additionally presented in the online supplemental tables S1 and S2.

Table 1

UK Biobank participants characteristics by current night shift work exposure (n=130 853)

Night shift work and risk of PCa

In the analysis of the current work schedule, model 1 showed that, compared with day workers, those who worked shifts but no night shifts (HR 0.83, 95% CI 0.74 to 0.92), some night shifts (HR 0.82, 95% CI 0.72 to 0.92) and usual night shifts (HR 0.71, 95% CI 0.61 to 0.82) had a reduced risk of PCa, with the risk decreasing as the frequency of night shifts increased (p for trend <0.001). However, after adjusting for confounding factors, this association was no longer significant. In model 2, no significant association was found between shift work and PCa risk, whether it involved shifts but no night shifts (HR 0.93, 95% CI 0.84 to 1.04), some night shifts (HR 1.07, 95% CI 0.95 to 1.21) or usual night shifts (HR 0.97, 95% CI 0.83 to 1.12). Model 3 was similar to those from model 2 (shift but no night shifts: HR 0.96, 95% CI 0.85 to 1.08; some night shifts: HR 1.16, 95% CI 0.99 to 1.33; usual night shifts: HR 1.01, 95% CI 0.85 to 1.19). In the analysis of the average frequency of night shift work, model 1 showed no significant impact of different night shift frequencies (<3/month: HR 1.03, 95% CI 0.86 to 1.25; 3–8/month: HR 0.90, 95% CI 0.80 to 1.01; >8/month: HR 0.90, 95% CI 0.79 to 1.02) on the risk of PCa. After adjusting for confounders in model 2 and model 3, the results did not change substantially. Additionally, an analysis of lifetime duration of night shift work was conducted. In model 1, a cumulative duration of <10 years of night shift work was associated with a reduced risk of PCa (HR 0.86, 95% CI 0.75 to 0.99), while ≥10 years of night shift work showed no significant association with PCa risk (HR 0.96, 95% CI 0.87 to 1.06). In model 2, no significant association was found for either <10 years (HR 0.92, 95% CI 0.80 to 1.05) or ≥10 years (HR 0.98, 95% CI 0.88 to 1.08) of night shift work. Similarly, in model 3, neither <10 years (HR 0.89, 95% CI 0.72 to 1.09) nor ≥10 years (HR 1.00, 95% CI 0.86 to 1.16) was associated with PCa risk (table 2). The results were similar in stratified analysis of age, chronotype, ethnicity, BMI and family history of PCa (online supplemental table S3). In the sensitivity analysis, we still did not observe any statistically significant results (online supplemental table S4).

Table 2

Night shift work and prostate cancer risk in the UK Biobank

Genetically predicted night shift work on PCa in MR

We conducted a comprehensive causal analysis, examining the causal relationship between night shift work and the risk of PCa across nine dimensions. The instrumental variables, represented by selected single nucleotide polymorphisms (SNPs), and the evaluation of night shift work across these dimensions are meticulously outlined in the supplementary materials (online supplemental methods and table S5). Our principal observational method, inverse variance weighting (IVW), did not reveal any discernible causal link between night shift work and PCa (table 3). Consistent findings were observed across various models in alignment with the IVW approach (table 3). Subsequent sensitivity analyses were conducted, and the leave-one-out analysis demonstrated that excluding any individual SNP did not induce alterations in our results (online supplemental figure S2). In the examination of SNP heterogeneity and pleiotropy, only the SNP corresponding to ‘Night shifts worked: Shift pattern was worked for the whole of the job’ exhibited heterogeneity (IVW p=0.021, MR-Egger p=0.037), while the remaining SNPs showed no evidence of heterogeneity or pleiotropy (online supplemental table S6).

Table 3

Findings of summary-level MR investigating effects of night shift work on PCa

Discussion

In the present investigation, we used a large-scale cohort study methodology based on the UKB and implemented two-sample MR analysis to elucidate the influence of night shift work on PCa. Nevertheless, our research outcomes reveal that, irrespective of the study approach employed—be it cohort studies or MR analysis—no substantial association between night shift work and the risk of PCa seems to exist.

It is indisputable that night shift work is associated with well-defined carcinogenic mechanisms. Current studies on the carcinogenic mechanisms of night shifts mainly include the following: (1) exposure to light at night, which suppresses the nocturnal peak of melatonin and its associated anticancer effects; (2) regulation of clock genes controlling apoptosis and cell proliferation was disrupted; (3) internal desynchronisation and regulatory defects of cellular circadian cycles; (4) sleep deprivation alters immune function19; (5) leads to disturbance of sex hormone secretion.28 Among them, melatonin may play the most important role. Studies have found that melatonin has a good anti-cancer effect, which is specifically manifested in (1) the reduction of endogenous melatonin leads to immunosuppression, which may affect the development and growth of cancer cells; (2) endogenous melatonin itself has a tumour suppressor effect; (3) protect cells from DNA damage and promote the repair after DNA damage.29 In studies related to PCa, Lozano-Lorca et al found that melatonin levels in patients with PCa were significantly lower than in men without PCa.30 Another study reported a negative correlation between first morning urinary melatonin-sulfate levels, the melatonin/cortisol ratio and both the incidence of overall PCa and advanced-stage PCa.31 These findings suggest a link between night shift work, which disrupts circadian rhythms, and the development of PCa.

Nevertheless, there remains a debate regarding the potential association between night shift work and the incidence of PCa, although there have been a series of studies conducted on the association between night shifts and PCa. On comprehensive examination of pertinent studies, it was observed that several case-control studies indicated an elevated risk of PCa associated with night shift work.8–16 Conversely, the majority of prospective cohort studies reported no significant association between night shift work and PCa.17–23 For instance, a case-control study conducted in Spain, encompassing 1095 PCa cases and 1388 randomly selected population controls, revealed an increased risk of PCa with prolonged night shift work (≥28 years: OR 1.37, 95% CI 1.05 to 1.81).17 In contrast, a cohort study by Hammer et al, involving 27 828 individuals, found no heightened risk of PCa among night shift workers compared with their daytime counterparts (HR 0.93, 95% CI 0.73 to 1.18).9 Moreover, a comprehensive occupational risk survey covering 15 million individuals across Nordic countries failed to identify any occupational risks associated with PCa.16 In a recent meta-analysis, researchers systematically reviewed eight cohort studies, yielding a pooled HR of 1.0 (95% CI 0.6 to 1.7). The analysis results demonstrate a non-association between night shift work and PCa.32 These findings are consistent with the outcomes of our analysis.

Different definitions and measurement criteria for night shift work may be the main reason for the disparate results. Alexander et al, in their meta-analysis, underscored the substantial inconsistency in defining night shift work across existing cohort studies examining the association with PCa risk. Moreover, significant disparities were noted in the adjustment of pertinent confounding variables among different studies, introducing a considerable risk of bias and diminishing the overall reliability of research findings.32 Our study used the definition of night-shift work as stated by the IARC, which involves working at least 3 hours between midnight and 05:00.23 We also recommend the use of IARC recommendations in other related studies to ensure comparability of subsequent studies. Furthermore, the nature of different types of night shift work may also contribute to the variability in research findings. It is foreseeable that various night shift jobs differ significantly in terms of work intensity, environment and nighttime light exposure. However, most current studies do not provide detailed descriptions of the specific occupations of the participants.

To our knowledge, this represents the inaugural study employing MR to ascertain the causal relationship between night shift work and PCa. It also stands as one of the few investigations offering a comprehensive evaluation of the impact of night shift work on PCa, considering current work schedule, frequency of night shift work, and lifetime duration of night shift work. Ultimately, our study affirms the lack of influence of night shift work on the heightened risk of PCa. Nevertheless, inherent limitations are acknowledged. Primarily, despite diligent attempts to assess the extent of night shift work from multiple dimensions, certain aspects remain unaccounted for. Given the intricate nature of night shift work, a 2009 workshop organised by the IARC convened experts to standardise essential metrics for epidemiological studies on night shift work. These metrics encompass night shift work hours (hours/week), the definition of night shifts, duration (years of non-day shift work), intensity of night shifts (number of night shifts per month/year), cumulative night shift exposure, type of night shift rotation, rotation direction and speed, as well as post-night shift sleep conditions and nighttime light exposure.33 Jennifer et al conducted a meta-analysis of nine observational studies. The findings demonstrated a statistically significant elevation in the risk of PCa among rotating night shift workers compared with their daytime counterparts, while no such association was observed within the fixed night shift group.34 It is apparent that several unaccounted factors may have implications for the precision of our research findings. Furthermore, the night shift exposure data employed in this study are exclusively derived from questionnaire surveys, introducing an inherent potential for information bias. Previous studies have confirmed that night shift work can increase the risk of aggressive PCa.21 In this study, however, due to the lack of pathological data, it was not possible to conduct subgroup analyses based on PCa subtypes, which may obscure certain true associations.

Conclusion

In light of our research outcomes, our investigation did not find any association between night shift work and PCa. However, as elucidated by Kyriaki et al, night shift work constitutes a highly intricate exposure variable necessitating comprehensive evaluation from various dimensions. The prevailing cohort studies in this domain predominantly rely on suboptimal assessments of night shift exposure, incomplete work histories and abbreviated cancer follow-up duration, thereby compromising the credibility of many investigations.35 Consequently, this study marks a point of departure rather than a culmination in our research endeavours. In future research, we will further explore various factors, including shift patterns, different types of night shift work and night light exposure caused by night shifts. Additionally, we will conduct an in-depth investigation into different subtypes of PCa.

Ethics statementsPatient consent for publication

Not applicable.

Ethics approval

The UKB’s research ethics committee and Human Tissue Authority research tissue bank approvals mean that researchers wishing to use the resource do not need separate ethics approvals (unless recontact with participants is required). All the participants provided informed consent, exempted this study. Participants gave informed consent to participate in the study before taking part.

Acknowledgments

We express our gratitude to the research communities and individual researchers for their invaluable contributions in making Genome-Wide Association Study (GWAS) summary data publicly accessible. Our sincere appreciation goes to the IEU open GWAS project, UK Biobank and the Finngen consortium for generously providing the summary-level data used in this study. Special thanks to all researchers who have contributed to our research endeavour and enriched the scientific discourse in this field.