Association Between Obstructive Sleep Apnea and Cancer: A Survival Analysis

Epidemiology studies on cancer risk attributable to obstructive sleep apnea (OSA) are inconsistent. There may exist differences in cancer risk when assessing OSA severity by the Apnea-Hypopnea Index (AHI) versus time spent below 90% oxygen saturation (T90%). We examined a clinical population undergoing polysomnography (PSG) to determine the impact of OSA severity on cancer risk by AHI and T90%. We collected records from patients undergoing PSG between 1998-2018 and linked them with electronic medical records to determine time-to-cancer diagnosis following PSG. We constructed a multivariable Cox proportional-hazards model to measure the impact of OSA severity, determined by AHI or T90%, on cancer incidence that adjusted for age, sex, BMI, hypertension, chronic obstructive pulmonary disease, coronary artery disease, diabetes, center, race, and ethnicity. We then used this model to assess the impact of OSA severity on risk for specific cancers. As cancer risk is differential between the sexes, we also stratified our analysis by sex. Among, 37,998 patients undergoing polysomnography, 3,218 cancers were diagnosed during follow-up (median: 5.9 years, interquartile range: 2.4-8.6). AHI was not significantly associated with an increase in cancer risk after adjusting for confounders (Hazard Ratio (HR) 0.93; 95% CI, 0.84 – 1.04; P = 0.22). Sex stratified analyses did not show a strong association between AHI severity and cancer risk. High T90% (T90%>13.3) was found to confer an 18% increase in cancer risk (HR 1.18; 95% CI, 1.05 – 1.33; P = 0.0069) compared to those with low T90%. Among specific malignancies, bladder and lymph/hematological cancer risks were significantly elevated for those with high T90%. The association between overall cancer risk and high T90% did not vary significantly by sex (P-interaction = 0.20). High T90% was associated with a significant increase in overall cancer among men (HR 1.22; 95% CI 1.02 – 1.46; P = 0.029), particularly bladder and lymph/hematological cancers. Our results have public health significance because they suggest that patients who experience nocturnal hypoxia have elevated cancer risk, and that the increased cancer risk may not be identified AHI alone. Future studies warranted that emphasize both T90% and AHI in determining cancer risk in patients with OSA