The incidence of obstructive sleep apnea (OSA) is increasing worldwide, affecting approximately 15% of males and 5% of females in the North American population (1). In South Korea, the prevalence of OSA is reportedly 4.5% in males and 3.2% in females (2). Patients with OSA show intermittent hypoxemia (IH) with or without hypercapnia, resulting in functional changes in the autonomous nerve system (3–5), as well as structural or molecular changes in the cardiovascular (6), neurologic (7), immune (8), or endocrine organs (9). Therefore, OSA acts as an independent predisposing factor for developing various cerebrovascular and metabolic disorders (10). One additional aspect of managing OSA lies in its efficacy in reducing the risk of the development of, for example, ischemic heart disease, stroke, and type 2 diabetes mellitus (DM) with decreased morbidity (11, 12).

Interestingly, it has been claimed that OSA may be positively associated with various malignant neoplasms in humans (13). Justeau et al. (14) reported that patients with more severe nocturnal IH who were untreated showed a significant increased risk in all-cancer incidence independently; of all the cancers, lung cancer showed the most statistical significance.

Lung cancer stands as the leading cause of cancer-related deaths in men and second-greatest cause in women; approximately 2.1 million patients were diagnosed, and 1.8 million deaths were recorded worldwide in 2018 (15). In South Korea, the crude incidence of lung cancer reported in 2019 was 58.4 per 100,000, affecting males and females at rates of 79.4% and 37.4%, respectively (16). Although the exact mechanism through which OSA develops into lung cancer is uncertain, the possible carcinogenetic and cancer-progression potential of OSA in lung cancer was reported in the previous study (17).

To date, numerous meta-analysis studies regarding OSA and cancer incidence have been published, suggesting that lung cancer is one of the cancers correlated with OSA (12, 17–20). Nevertheless, a study focusing on the incidence of newly diagnosed lung cancer in patients diagnosed with OSA in a large, nation-wide cohort has not yet been published to the authors’ knowledge, particularly in the East Asian population. Additionally, no previous studies have conducted a subgroup analysis on the impact of gender differences regarding OSA and lung cancer development.

The authors sought to determine the incidence and hazard ratio of lung cancer in patients with OSA using a large, nationwide retrospective cohort stretching twelve years, utilizing data from the Korea National Health Insurance Service (KNHIS) database. The study incorporated analyses of subgroups based on gender and age.

According to the KNHIS records, the total number of registered individuals in 2007 was 49,570,064. Throughout the patient recruitment phase ranging from January 2007 to December 2017, a total of 310,557 patients were registered with newly diagnosed OSA, as depicted in Figure 1 . The control group comprised a total of 1,339,245 individuals who were recruited. During the observation period spanning from January 2007 to December 2019, the retrospective cohort’s mean follow-up time interval was 5.9 ± 3.1 years, as determined by the calculated mean and standard deviation. A p-value below.05 was deemed statistically significant.

The results of this current study, which were obtained from the nation-wide retrospective cohort from 2007 to 2017 with a 12-year observation period (2007 to 2019), showed that the adult patients diagnosed with OSA had a significantly reduced risk of developing lung cancer. This reduced risk was significantly valid when the cofounding demographic variables were adjusted. However, in the subgroup analysis according to each gender, only the male OSA patients showed a significantly reduced adjusted HR of 0.84 (95% CI; 0.78–0.90), whereas the adjusted HR of the females did not show such statis-tical significance. Despite potential limitations such as selection bias and inadequate adjustment for various lung cancer risk factors, our analysis of the KHNIS dataset suggests that male patients with obstructive sleep apnea have a lower risk of developing lung cancer. Gender disparity in the carcinogenesis pathway of lung cancer may exist, as females may not exhibit the same pattern.

The strength of our study lies in its large statistical power, owing to the fact that all South Korean citizens are mandatorily covered by the National Health Insurance Service. In addition, the observation of both the OSA and the control group for twelve consecutive years in a retrospective manner may strongly support the causal association of the carcinogenic potential of OSA with lung cancer in our results when compared with cross-sectional or case-control studies. Moreover, the application of a subgroup analysis of the lung cancer risk of OSA according to the gender and age-group differences in the large retrospective cohort in this study was not reported in previous studies. To the authors’ knowledge, this study provides the largest recorded power and causality for revealing the incidence of lung malignancies in OSA patients.

Recently, sleep-related breathing disorders (SBD), including OSA, as independent risk factors for developing various malignant tumors in the human body, have gained interest from many researchers (13, 14, 22). The chronic IH in OSA followed by various responses in the neuroendocrine, cardiovascular, and respiratory organs may facilitate tumor growth and progression in various organs (23). Many studies from various nations have suggested OSA as an independent risk factor for all-cancer incidence. For instance, the relative risk was 1.26 in Sillah et al.’s study (24), and three meta-analyses resulted in all-cancer-risk values of 1.49, 1.53, and 1.52. Additionally, previous results showed dose–response relationships between OSA and cancer incidence. Palamaner Subash Shantha et al. (18) reported that overall cancer incidence and mortality were higher by up to three times in severe OSA than in mild-to-moderate OSA. However, some studies presented results that opposed this finding, suggesting instead that OSA was not a significant risk factor in the development of all types of cancer (25, 26). Some cancers are more likely to be correlated with OSA, while other types of malignant tumors are not (14, 20). In the previous analyses, in terms of cancers of various organs, colorectal (27), breast (28), prostate (9), pancreas (29), and renal-cell carcinomas (29) were shown to be significantly affected by OSA, as OSA served as a significant risk factor for developing these malignancies. Together, the outcomes of previous clinical and epidemiological studies suggest the potential clinical impact of OSA on the incidences of various cancers.

Contradictory results were obtained regarding whether OSA may affect lung cancer incidence and mortality in previous studies. In some cross-sectional studies with relatively large numbers of subjects, the incidence of SBD or OSA reached 49% to 80% in all patients newly diagnosed with lung cancer, and approximately half of the lung cancer patients had moderate-to-severe OSA (30–33). Huang et al. (32) reported that stage III and IV advanced-lung cancer patients with an apnea–hypopnea index (AHI) greater than 30 had a three-year mortality rate of 80%, which was reduced to 25% in patients with an AHI less than 15. This study strongly connected the effect of the IH of OSA with carcinogenesis and tumor progression in lung cancer. A large-scale cohort followed for a further 5 years and population-based studies further support OSA as an independent risk factor for higher lung cancer incidence. Huang et al. (32) reported an HR of 1.52 (95% CI, 1.07–2.16), and Jara et al. (34) reported an HR of 1.32 (95% CI, 1.27–1.37), showing a significantly in-creased HR of OSA in lung cancer incidence. Nonetheless, Kendzerska et al. (35) reported an HR of 1.38 (95% CI, 0.94–2.03) showing no statistical significance; insignificance was also reported in Gozal et al.’s study (29). Furthermore, Sillah et al. (24) reported that OSA was associated with a significantly reduced lung cancer incidence, as their HR value was 0.66 (95% CI, 0.54-0.79). The disagreement in the previous studies suggests the need for a well-designed, large, nation-wide cohort study with a relatively long duration of observation. The 12-year retrospective nation-wide cohort in our study design might explain and further support the discordant results on whether OSA might play a role in lung cancer development.

Our results demonstrate that the HR for developing new-onset lung cancer in OSA patients compared with the control group with different ages, sexes, incomes, and various comorbid diseases was 0.91 (95% CI; 0.85–0.97), showing a significantly reduced risk of developing lung cancer. An even greater risk reduction was observed in a model in which cofounding variables were considered, showing a significantly adjusted HR of 0.87 (95% CI; 0.82–0.93). Moreover, the cumulative incidence plot showed a significant difference between the OSA and control groups in the twelve-year retrospective cohort. The results indicate that OSA may not be associated with lung cancer development in the general adult population, regardless of other cofounding clinical variables. Our results provide a remarkable addition to the knowledge on this subject, as previous researchers disagreed over whether OSA increases the risk of lung cancer development.

In our study, there was a significant reduced risk of lung cancer in OSA individuals. Like our results, Sillah et al. s’ report where the OSA showed as a significantly protective agent for developing lung cancer (24). In Siallah et al. s’ study, all the included subjects were aged over 64 years. Similarly, our study subgroup analysis showed a significantly reduced risk only in the senile age (age>65) group, which carefully suggests a rather protective effect of OSA for lung cancer development in the geriatric population, but not in the young and mid-aged adult population. Moreover, Christensen et al. reported an increased risk of lung cancer in OSA patients in the patients under 50 years old (25), and an animal study on an intermittent hypoxic model revealed rather a protective effect for lung cancer development in the aged group, which the younger group did not (36). The impact of age on the increased susceptibility to OSA in lung cancer development should be more clarified and discovered in the future studies.

Theoretically, the sleep fragmentation (SF) and IH observed in OSA patients are two major features that could explain the increased cancer incidence in the OSA population. Although the exact mechanism of IH in carcinogenesis is not yet understood, the frequent hypoxia followed by normoxia in IH mimics a condition similar to the reperfusion injury in tissues that undergo ischemic stress (37). The production of reactive oxygen species by the endothelial cells of vascular structures exposed to chronic IH may predispose patients to carcinogenesis in normal tissues (38). Moreover, IH may up-regulate various hypoxia-inducible factors (HIFs) in many organs, which may promote carcinogenesis (39, 40). On the other hand, SF has been shown to result in activated sympathetic tone, chronic inflammation, and altered immune cell functions, all of which may promote carcinogenesis in various organs (41, 42). Furthermore, it has been also suggested that nasal obstruction, which is a common accompanying findings in OSA patients might also play a role attributing in chronic inflammation and oxidative stress, which all together lead to carcinogenesis (43).

Owing to the fact that OSA is caused by a narrowing of the upper airway in most cases, the lower airway tract and the lung are anatomically the principal areas affected by the sustained decrease in oxygen concentration following hypoventilation in OSA patients (44). Thus, the carcinogenic potential of IH in lower airway tract cells in OSA patients is a field of interest for many researchers (45). The chronic inflammation as a result of IH may synergize with various known inhalant carcinogens, such as tobacco smoking, in the lower airway, thus facilitating the formation of a microenvironment suited to lung cancer development (46). Another possible mechanism for lung cancer development and progression in OSA is via HIFs, as HIF-2α has shown potential for carcinogenesis, angiogenesis, and metastasis in lung cancer in various studies (38–40, 45, 47). Furthermore, Wnt/β-catenin signaling, which is thought to be enhanced through interaction with up-regulated HIF pathways, has been shown to activate the oncogenic potential in non-small-cell carcinomas of the lung (48).

In addition, managing OSA has been shown to reduce oxidative stress, free radicals, and inflammation, all of which can reduce the risk of carcinogenesis. Oxidative stress is an imbalance between the production of reactive oxygen species (ROS) and the ability of antioxidants to neutralize them. Multiple forms of cancer, including laryngeal cancer, have been associated with its presence (49, 50). Due to intermittent hypoxia and reoxygenation during sleep, obstructive sleep apnea syndrome (OSAS) has also been correlated with oxidative stress. OSAS may contribute to the development of laryngeal cancer by fostering chronic inflammation and oxidative stress, which can cause DNA damage and promote tumor growth (49). Moreover, OSAS may impede immune function, making it more difficult for the body to combat cancer cells.

Furthermore, it has been suggested that disruptions in the human circadian rhythm potentially lead to carcinogenesis in various organs (51); this includes lung cancer, which is one of the cancers related to disrupted circadian rhythms (52). As suggested in Koritala et al.’s article, various biomarkers and molecular changes indicating a disrupted circadian rhythm are observed in OSA patients, associated with sleep fragmentation and sleep arousal (53). Melatonin, an endogenous molecule secreted in the pineal gland and regulated by the circadian rhythm (54), is reportedly insufficiently secreted in OSA patients (55–57). Interestingly, many scholars have suggested the defensive role of melatonin in the carcinogenesis pathway (58, 59), and the decreased secretion of melatonin in OSA patients may lead to increased cancer risk (60). Taken together, the results of previous cellular and animal studies offer a molecular basis for further support for the oncogenic potential of OSA, including carcinomas of the lung.

Interestingly, significant lung cancer risk reduction was seen only in the males, and not in the females, regardless of the patients’ age group. The exclusive significance of female susceptibility to lung cancer after OSA highlights the novelty of our study. Wu et al. (20) reported an increased risk of OSA for all-cancer development, especially in the female population. However, in the previous literature, it was not clarified whether OSA increases cancer risk predominantly in females; for instance, Cheong et al. (17) raised this issue as a limitation in their meta-analysis. Previous studies, including some meta-analyses, reported a significantly increased risk of OSA with lung cancer incidence (17, 32, 34), while some studies did not (20, 29), and others even reported a significantly decreased risk (24). These contradictory results are thought to result from differences between the demo-graphic characteristics in each study. Huang et al.’s study (32) was based on a female-only cohort, which showed an HR of 1.52, while Jara et al.’s study (34) was mostly (94%) on male patients, resulting in an HR of 1.32 for lung cancer incidence in OSA patients, showing a significantly increased lung cancer incidence in both genders. However, it must be noted that both studies were derived from predominantly white and relatively homogenous occupational groups, as Huang et al.’s study (32) consisted of nurses and Jara et al.’s study (34) consisted of veterans. Thus, the conclusions derived from their results might be limited for the general population. Furthermore, it should be stated that most of the previous studies were published in Western countries, in which those of East Asian heritage form only a small proportion of the population.

The lung cancer epidemiology in East Asian populations showed a distinctive pat-tern compared with the Caucasian population; in South Korea, 36% were never-smokers, and more than up 70% of the never-smoking lung cancer patients were female and were histologically diagnosed with adenocarcinoma-expressing epidermal growth factor receptor (EGFR) mutations (61). Interestingly, Marhuenda et al. (62) reported different tumor proliferation patterns, expression levels of epithelial cell adhesion molecule, and different amounts of HIF-1α nuclear translocation between squamous cell carcinoma and EGFR mutation-positive and -negative adenocarcinomas derived from human pulmonary cell lines.

The exclusive increase in lung cancer incidence in the female population regardless of age might aid in the addition of new knowledge. Since our results showed the susceptibility of female patients with OSA specifically to developing lung cancer, the question of whether pulmonary tissues from never-smoking females are more vulnerable to IH in developing adenocarcinoma with EGFR mutations would be an interesting investigation topic in the future. An increased HR of 1.52 in Huang et al.s’ study which only consists of female (32), shows a higher HR compared with HR in Jara et al.s’ study, showing a HR of 1.32, which their study group only included male patients (34). The results of these two cohort studies goes in line with our results, suggesting an increased susceptibility to lung cancer development in the female OSA than the male OSA. Nonetheless, these two studies are separate, two independent studies, and there are no well-designed systemic study investigating on the gender difference in the incidence of lung cancer development in OSA patients. Therefore, we suggest that this gender difference in the incidence of lung cancer in OSA group might be an interesting topic in the future research.

Although our study presented some unprecedented findings with a twelve-year retrospective cohort on a nation-wide scale, we acknowledge some limitations that must be declared. First, we were not able to consider the social history, such as smoking and alcohol abuse, or familial history of lung cancer, which are some of the known major risk factors for lung cancer (63), in each patient, as these demographics were not included in the KNHIS data. Although we acquired data on the presence of COPD diagnosis, which is known to occur at a high rate in heavy tobacco smokers, we declare that these data might not quite represent the effect of tobacco smoking in our study. Furthermore, the study sub-jects’ nutritional status (i.e., body mass index), which is known to be associated with OSA, was not evaluated, as this information is not available in the KNHIS registry. Second, we were not able to provide a subgroup analysis on the subtypes of lung cancer in terms of the histological diagnoses or genetic profiles of lung cancer among the study subjects. Third, the KNHIS data only provided patients diagnosed with OSA, with no details regarding the OSA severity (represented as apnea-hypopnea index, etc.), obesity degree represented as body mass index, or whether the patients underwent treatment for OSA. Therefore, we were not able to analyze the dose-response of OSA severity and confounding impact of obesity in lung cancer development in OSA individuals. Along with the severity of OSA, BMI shall also be considered to elucidate this matter in the future study. Fourth, our research classified adult population as those over 20, whereas others have characterized it as 18 to 19. Consequently, the disparity in adult age criteria may limit the comparability of OSA’s effect on lung cancers with other studies. Last, the control group was only set to have 5 times larger number than the study group, and only matched with age and gender. These might possess the possibility of selection bias, thereby limit the results of our study. Furthermore. it should be stated that there is the possibility of an undiagnosed OSA population in the ‘control’ group in our cohort, as the OSA group in our cohort consisted of patients who received medical care and were confirmed to have OSA by a physician.

Adults with OSA had a slightly lower hazard ratio (0.87) for lung cancer development in the Korean population. The risk reduction was observed only in male OSA patients in the subgroup analysis by gender. Our study is the first study in the literature to raise the gender differences in the OSA impact in lung cancer development, analyzed with the largest number of retrospect cohort of 12 years on a national-wide scale.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

This study was approved by the Institutional Review Board (IRB) of Inha University Hospital (IRB no. 2022-11-004). Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements.

Conceptualization: MP, JCho, and JChoi. Formal analysis: K-DH, JCho, and JChoi. Investigation: K-DH, JCho, and JChoi. Methodology: K-DH, JCho, and JChoi. Resources: JCho and JChoi. Writing–original draft: MP, JCho, and JChoi. Writing–reviewing and editing: MP, JCho, and JChoi. All authors contributed to the article and approved the submitted version.