Sex-specific associations between habitual snoring and cancer prevalence: insights from a US Cohort Study

Methods

Statistical Analysis

In our observational cross-sectional analysis, participants were categorized by snoring status, and subject characteristics were compared using standardized mean differences. We summarized rates of cancer, any and by subtypes, sex, and snoring frequency.

To balance demographic, clinical, and behavioral risk factors, separate propensity score models were constructed for men and women, with snoring status as the outcome and participant characteristics as predictors. To account for missing covariates, we used multiple imputations with 20 imputations, constructing a separate logistic regression model on each imputed dataset. The propensity model included all demographic and clinical factors in Table 1. Recognizing OSA as a potential confounder due to its association with increased cancer risk and hypoxic burden [23–25], we included OSA as a variable in our propensity model to help balance this confounder across snorers and non-snorers. Predicted probabilities were then used as the inverse probability of treatment weights (IPTW) in subsequent analyses, with analyses performed separately on each imputation set and the results pooled.

For hypothesis testing, IPTW logistic regression models were employed to evaluate cancer risk by snoring status, calculating odds ratios (ORs) with 95% confidence intervals (95% CIs) for each sex and cancer type. Subgroup analyses were conducted based on sleep duration (short, long, and average), age (<50, 50–64, and 65 years and above), and BMI categories (underweight, healthy, overweight, and obese) to assess the consistency of associations. Additionally, dose–response relationships were explored by analyzing cancer risk across frequencies of snoring (never, 1–2, 3–4, or 5–7 nights a week). All analyses were performed using Stata version 18 (StataCorp, College Station TX).

Results

Discussion

Our retrospective cohort study reveals significant sex disparities in the relationship between habitual snoring and cancer prevalence. Using IPTW-weighted analysis to balance demographic, clinical, lifestyle, and sleep-related characteristics, we observed a 26% reduction in overall cancer odds among men with habitual snoring, particularly in lung, colon and rectum, skin, melanoma, and leukemia subtypes. Conversely, among women, no significant difference was observed in overall cancer prevalence except for breast and cervix cancers. Drawing from a dataset of 15 892 nationally representative adults, we also found age played a crucial role in habitual snoring-cancer relationships, with inverse associations noted in men aged ≤49 years and ≥65 years, and positive associations in young and middle-aged women. Sleep duration further influenced these associations, with shorter duration (<7 hours) linked to lower overall and lung cancer prevalence among snoring men, and longer duration (>9 hours) associated with higher overall and uterine cancer prevalence among snoring women. These nuanced findings underscore the complex interplay between habitual snoring and cancer risk across different life stages, sleep durations, and BMI subgroups, emphasizing the need for individualized evaluation and consideration of sex, age, and sleep-related factors in future mechanistic research.

Our study represents the first to document a significant inverse association between habitual snoring and cancer prevalence in men, particularly among males aged ≥65 years or with short sleep duration. While we adjusted for OSA diagnosis, our study, akin to prior epidemiologic research, cannot discern the impact of simple snoring versus snoring with OSA on cancer prevalence. Moreover, underdiagnosis of OSA is prevalent in the general population, with an estimated 93% of women and 82% of men with moderate-to-severe OSA being unaware of their condition [26, 27]. Existing studies on the potential link between OSA and cancer exhibit inconsistency, and sex-specific aspects of this association remain poorly studied [28]. For instance, a study of 267 849 individuals from the Korea National Health Insurance Service registry indicated a lower risk of lung cancer development in males with OSA but not in females, especially those aged>65 years [29]. However, the study’s generalizability was limited by its failure to account for major behavioral and clinical lung cancer risk factors and its homogenous sample [29]. Our findings extend these insights by incorporating demographic, lifestyle, clinical, and sleep-related factors into the analyses, exploring snoring-related cancer prevalence across diverse patient populations. Additionally, an analysis involving 4580 individuals aged 65 and above from the Cardiovascular Health Study highlighted an inverse association between snoring and incident overall cancer risk, considering gender, age, BMI, diabetes, physical activity, and alcohol use [30]. This analysis suggests that the type of cancer and age might influence the epidemiological association with sleep problems and that older age could confer protection against intermittent hypoxia-induced carcinoma growth [30, 31]. These assumptions align with our observed inverse association between habitual snoring and overall cancer, colon and rectum, skin, melanoma cancer in males aged 65 and above, and breast cancer in women aged 65 and above.

A surprising finding of this study was the lower cancer prevalence among habitual snorers compared to non-habitual snorers, particularly in men. While snoring is widely recognized as a primary symptom of OSA [32, 33]—a condition linked to increased cancer risk [8, 23–25, 34]—our findings suggest a more complex relationship. Previous studies have indicated that the association between OSA and cancer typically emerges in more severe cases of apnea, as evidenced by elevated cancer risks in patients with moderate to severe OSA compared to those with mild OSA [25]. Hypoxia and sleep fragmentation are considered to be the main pathologic link in the OSA-cancer incidence relationship [35, 36]. In murine-based reports implicating sleep fragmentation and intermittent hypoxia components of OSA, fragmented sleep accelerates tumor growth and progression through tumor-associated macrophage recruitment and TLR4 signaling [35]. On the other hand, hypoxia increases angiogenesis into the tumoral tissue and induces overexpression of transcription factors, such as hypoxia-inducible factor-1, which are known to trigger upregulation of proangiogenic mediators, such as vascular endothelial growth factor in tumor cells, and enhance tumor progression [36, 37].

Notably, snoring alone does not necessarily indicate the presence of OSA [33]. Advanced signal processing algorithms-based studies have demonstrated only a weak positive correlation between snoring frequency and the apnea–hypopnea index (AHI), a standard measure of OSA severity [33]. This suggests that snoring without accompanying severe apnea may not significantly elevate cancer risk, a hypothesis supported by our data and other population-based studies [38]. For instance, Gozal et al. identified selective cancer risks associated with OSA using a nationwide employee-sponsored health insurance database, where certain cancers like pancreatic and kidney cancers showed increased risk, whereas others like colorectal and breast cancers demonstrated lower risks among those diagnosed with OSA [38]. This aligns with our findings of reduced colorectal and breast cancer prevalence in men and female habitual snorers, respectively, potentially highlighting a complex interplay of factors beyond simple hypoxia-induced carcinogenesis.

Our study refines this perspective by incorporating IPTW to control for important confounders such as smoking, BMI, race/ethnicity, and other demographic characteristics, unlike previous studies that might not have fully accounted for these variables [38]. Despite this, the pathophysiological mechanisms underlying the reduced risk of certain cancers in snorers remain speculative. It is hypothesized that varying cancer types may interact differently with hypoxia-inducible factors, which play critical roles in tumor progression and response to hypoxia [38–41]. Given these complexities, the findings necessitate cautious interpretation, particularly due to the cross-sectional design of our study which limits causal inferences. Future research should aim to replicate these results in larger, prospective cohorts with objective measures of snoring and OSA severity, alongside detailed cancer diagnoses, to better delineate the relationships and underlying mechanisms at play.

Sex differences in the association between habitual snoring and cancer may stem from distinct effects of carcinogenic risk factors in men and women. Genetic studies suggest that snoring correlates differently with lifestyle and clinical traits in each gender. For instance, while snoring prevalence is typically higher in males, tobacco use appears to have a stronger association with snoring in females, whereas alcohol consumption exerts a greater influence on snoring in males [13, 42]. Additionally, population-based research indicates higher cancer rates among heavy-drinking females compared to males [43, 44]. Moreover, several studies have shown a higher incidence of lung cancer in men than in women with similar tobacco smoking exposure levels [45, 46]. Although genetic evidence supports a causal relationship between BMI or whole-body fat mass and snoring, a prior review suggests a higher overall incidence of obesity-related cancer in females compared to males [13, 47]. Although no clinical studies have directly explored the sex-specific pathologic interaction between habitual snoring, carcinogenic risk factors, and cancer, these previous findings imply that varying carcinogenic potencies may contribute to specific cancer risk in individuals with habitual snoring, depending on gender.

Our findings contribute to the existing literature on the gender-specific influence of sleep duration on cancer risk. For instance, a recent study of 14 851 participants from the China Health and Retirement Longitudinal Study revealed elevated overall cancer risk among women sleeping less than 6 hours per night, but not in men [48]. Similarly, an analysis of 469 691 individuals in the UK BioBank identified both insufficient and excess sleep duration as independent risk factors for lung cancer, yet did not find an increased risk associated with snoring [49]. However, these studies did not separately investigate the interaction between sleep duration and snoring in cancer risk for men and women. Our weighted analysis, stratified by sleep duration, revealed that the inverse associations between habitual snoring and overall and lung cancer prevalence were observed only in males sleeping less than 7 hours. Conversely, in females, sleeping more than 9 hours per night was associated with increased odds of uterine and overall cancer prevalence. Furthermore, a systematic review and meta-analysis of cohort studies suggest that long sleep duration and sleep disturbance disrupt the immune-inflammation balance, leading to increases in systemic inflammation markers, which are associated with tumorigenesis [50, 51]. While long sleep might be a marker of chronic conditions [52], these findings support the potential role of chronic inflammation in cancer development, and suggest the possibility of using long sleep duration and habitual snoring as screening markers for cancer risk in women. While further research is warranted to explore the exact biological mechanisms, our study provides valuable insights into potential screening strategies for cancer risk.

There is ongoing debate about the impact of sleep duration on cancer risk. Recent analyses, including Mendelian randomization studies and systematic reviews, have further nuanced our understanding of the relationship between sleep duration and cancer risk [53–58]. For instance, Titova et al. (2021), using Mendelian randomization, found suggestive links between genetic predispositions to shorter sleep durations and increased risks of gastrointestinal cancers, although these findings did not withstand multiple testing corrections [53]. Similarly, a pooled analysis of Japanese cohorts indicated a non-linear association, where both very short and long durations of sleep correlated with an increased risk of certain cancers in men and women of certain age and BMI categories [54]. These findings align with our observations that both short and long sleep durations can modulate cancer risks differently across genders and cancer types, potentially through mechanisms involving systemic inflammation and immune function disruption [53, 54]. Notably, while some studies found significant associations for site-specific cancers, comprehensive reviews suggest that the overall cancer risk might not be significantly associated with sleep duration, highlighting the complexity of these relationships and the potential influence of unmeasured confounders [58]. Given these mixed outcomes, our study contributes to the ongoing discourse, suggesting that sleep duration, much like other lifestyle factors, potentially interacts with biological processes in a manner that may increase or decrease cancer risk. Our findings highlight the need to consider habitual snoring as a potential co-factor in these analyses. Future studies should aim to clarify these pathways and consider the impact of circadian rhythm disruptions, providing a more detailed landscape of how habitual behaviors like snoring and variations in sleep duration intersect with cancer epidemiology.

Previous studies have implicated intermittent hypoxia and sleep fragmentation as key mechanisms linking OSA to cancer incidence [28, 29, 59]. Intermittent hypoxia, characterized by repeated episodes of oxygen desaturation followed by reoxygenation, may lead to the generation of reactive oxygen species, similar to reperfusion injury in ischemic stress [60]. This process could predispose individuals to carcinogenesis by promoting reactive oxygen production in vascular endothelial cells [61]. Additionally, intermittent hypoxia may up-regulate hypoxia-inducible factors, altering substrate metabolism, angiogenesis, and cell differentiation, thus promoting cancer development [62]. Sleep fragmentation, on the other hand, has been associated with sympathetic activation, chronic systemic inflammation, and altered immune cell function, all of which may contribute to carcinogenesis in various organs [63, 64]. However, data on sex differences in the association between OSA and cancer are limited [28], and further research is needed to elucidate the potential role of cancer subtype, hormonal influences, and duration of OSA or habitual snoring on sex-specific mechanisms of carcinogenesis.

Limitations

Despite utilizing a large, diverse population sample and employing IPTW-weighted analysis were used in this study, several limitations must be acknowledged. First, sleep-related characteristics and cancer prevalence were self-reported rather than objectively measured, potentially introducing participant-specific recall bias and misclassification of exposure and response variables. While the definition of habitual snoring may misclassify cases where participants lack a bed partner into control, prior studies suggest high accuracy (up to 96%) of self-reported cancer diagnosis [65]. Secondly, our study design was observational, precluding the establishment of causality. Thirdly, the use of simple measures of sleep characteristics in a cross-sectional survey did not account for changes in sleep patterns over time, necessitating longitudinal cohort studies with repeated measures to validate findings. Additionally, our results may be susceptible to residual confounding from unaccounted covariates, such as medication use (e.g. chemotherapy), and genetic factors, which are not included in our dataset. Future studies should consider incorporating these clinical variables to more thoroughly explore the relationship between habitual snoring and cancer prevalence in men and women. Despite efforts to balance demographic, clinical, lifestyle, and sleep-related characteristics through weighted analyses in our main and subgroup analyses, residual confounding remains a possibility.

Conclusions

Our study elucidates sex-specific associations between habitual snoring and cancer prevalence, revealing inverse associations in males, particularly among males aged over 65 years, and varied effects in females, including lower odds of breast cancer but increased odds of cervix cancer. Additionally, we highlight the modifying role of sleep duration, with shorter duration (<7 hours) linked to decreased overall and lung cancer prevalence in snoring men, and longer duration (>9 hours) associated with elevated overall and uterine cancer prevalence in snoring women. Future research exploring the interactions between different sleep characteristics and cancer could have profound implications. Initiating long-term clinical trials to investigate cancer subtype screening based on objective sleep assessment would refine risk delineation, potentially altering prevention and screening strategies in a more accurate and individualized manner. Integration of genetic, circulating biomarkers, and other clinical and behavioral risk factors in future studies could provide insights into shared mechanisms underlying habitual snoring and cancer.

Supplementary Material

Author Contributions

Qinglan Ding (Conceptualization [equal], Data curation [lead], Investigation [lead], Methodology [equal], Project administration [lead], Resources [lead], Supervision [lead], Validation [equal], Writing—original draft [lead], Writing—review & editing [lead]), Jeph Herrin (Formal analysis [lead], Methodology [lead], Validation [lead], Visualization [lead], Writing—review & editing [equal]), and Meir Kryger (Conceptualization [lead], Investigation [equal], Methodology [equal], Resources [equal], Supervision [equal], Writing—review & editing [equal])

Data Availability Statement

The data for this study were sourced from NHANES, which is publicly available and can be accessed through the website https://wwwn.cdc.gov/nchs/nhanes/default.aspx.