Abstract
Background
Conventional metrics to evaluate sleep-disordered breathing (SDB) have many limitations, including their inability to identify subclinical markers of cardiovascular (CV) dysfunction.
Research Question
Does sleep study-derived circulation time (Ct) predict mortality, independent of CV risks and SDB severity?
Study Design and Methods
We derived average lung to finger Ct (LFCt) from sleep studies in older men enrolled in the multicenter Osteoporotic Fractures in Men (MrOS) Sleep study. LFCt was defined as the average time between end of scored respiratory events and nadir oxygen desaturations associated with those events. We calculated the hazard ratio (HRs) for the CV and all-cause mortality by LFCt quartiles, adjusting for the demographic characteristics, body habitus, baseline CV risk, and CV disease (CVD). Additional models included apnea-hypopnea index (AHI), time with oxygen saturation as measured by pulse oximetry (SpO2) < 90% (T90), and hypoxic burden. We also repeated analyses after excluding those with CVD at baseline.
Results
A total of 2,631 men (mean ± SD age, 76.4 ± 5.5 years) were included in this study. LFCt median (interquartile range) was 18 (15-22) s. During an average ± SD follow-up of 9.9 ± 3.5 years, 427 men (16%) and 1,205 men (46%) experienced CV death and all-cause death, respectively. In multivariate analysis, men in the fourth quartile of LFCt (22-52 s) had an HR of 1.36 (95% CI, 1.02-1.81) and 1.35 (95% CI, 1.14-1.60) for CV and all-cause mortality, respectively, when compared with men in the first quartile (4-15 s). The results were similar when additionally adjusting for AHI, T90, or hypoxic burden. Results were stronger among men with no history of CVD at baseline.
Interpretation
LFCt is associated with both CV and all-cause mortality in older men, independent of baseline CV burden and SDB metrics. LFCt may be a novel physiologic marker for subclinical CVD and adverse outcomes in patients with SDB.
Key Words: cardiovascular, circulation time, mortality, sleep apnea, sleep-disordered breathing
Abbreviations: AHI, apnea-hypopnea index; CSA, central sleep apnea; CSR, Cheyne-Stokes respiration; Ct, circulation time; CV, cardiovascular; CVD, CV disease; HF, heart failure; HR, hazard ratio; LFCt, lung to finger Ct; MrOS, Osteoporotic Fractures in Men; PSG, polysomnography; SDB, sleep-disordered breathing; SpO2, oxygen saturation as measured by pulse oximetry; T90, time spent with SpO2 < 90%
Sleep-disordered breathing (SDB) is a common condition that is increasingly recognized as one of the major risk factors for cardiovascular (CV) disease (CVD) and mortality.1, 2, 3, 4 Although conventional metrics focusing on the quantification of respiratory events are useful in the diagnosis of SDB, they may not optimally reflect the complex pathophysiologic consequences of SDB on the CV system.5,6 Despite extensive data obtained from polysomnography (PSG), only a handful of metrics, such as apnea-hypopnea index (AHI) and burden of hypoxemia, quantified by the total time spent with oxygen saturation as measured by pulse oximetry (SpO2) < 90% (T90), are typically used. There are a number of established as well as other physiologic variables that may be useful in risk assessment in patients with SDB. Sleep study-derived circulation time (Ct) is a physiologic measure that conceptually reflects oxygen transport time from lung to periphery and forms the basis for sleep study-based Ct measurement.7 Long Ct is a hallmark of circulation delay in patients with heart failure (HF).7,8 Therefore, prolonged Ct may be a novel marker for subclinical HF and may be associated with poor outcomes in patients with SDB. Therefore, in this study, we investigated whether prolonged Ct is associated with increased mortality in older adults with SDB.
Materials and Methods
Participants
We included participants of the Osteoporotic Fractures in Men (MrOS) Sleep study. The MrOS Sleep study (N = 3,135) is an ancillary study, Outcomes of Sleep Disorders in Older Men, of the parent MrOS cohort (N = 5,994) study that enrolled community-dwelling older men from six geographic regions in the United States: Birmingham, AL; Minneapolis, MN; Palo Alto, CA; the Monongahela Valley near Pittsburgh, PA; Portland, OR; and San Diego, CA (https://mrosonline.ucsf.edu/).9,10 Participants who reported regular use of CPAP at baseline were excluded.
Sleep study
An overnight unattended, in-home PSG test (Safiro; Compumedics) was conducted. A standard recording montage included EEGs (C3/A2 and C4/A1), bilateral electrooculograms, and a bipolar submental electromyogram; thoracic and abdominal respiratory inductance plethysmography; airflow (by nasal-oral thermocouple and nasal pressure cannula); finger pulse oximetry (Nonin Medical Inc.); lead I ECG; body position (mercury switch sensor); and bilateral tibialis leg movements (piezoelectric sensors). Apneas were defined as absence or near absence of airflow for ≥ 10 s regardless of desaturation. Hypopneas were defined as ≥ 30% reduction of breathing amplitude lasting ≥ 10 s. AHI was calculated based on the number of all apneas and hypopneas associated with a ≥ 3% desaturation or EEG-based arousal per hour of sleep. SDB was considered to be clinically significant if AHI was ≥ 15/h. Central sleep apnea (CSA) was determined if the central apnea index was > 5/h. Predominant CSA was determined if CSA was > 5/h and > 50% of AHI. Predominant CSA with Cheyne-Stokes respiration (CSR) was characterized by a minimum consecutive 10-min period of a crescendo-decrescendo respiratory pattern culminating in a nadir of central apnea with typical cycle length. T90 was derived and used as a conventional measure of nocturnal hypoxemia. In addition, the hypoxic burden, a predictor of mortality and incident HF in recent studies,4,11 was used as a novel measure of OSA-specific severity of hypoxemia. Data were scored by certified research sleep technologists at the central sleep reading center by using standardized criteria.12 For this study, only those who had at least 50 respiratory events (ie, apneas and hypopneas), allowing sufficient number of events for Ct measurement, were included.
Lung to Finger Ct
Ct was determined by measuring the time between initiation of hyperpnea (oxygen entry) after an apneic or hypopneic episode and the nadir point of SpO2, calculated from a subject-specific ensemble-averaged SpO2 signal.8 Because SpO2 was measured from pulse oximetry in the fingertip, we called it lung to finger Ct (LFCt).8
The LFCt was calculated from the aggregate (ensemble-averaged SpO2) number and not individual events. Figure 1 demonstrates the concept of LFCt by using a single event, and Figure 2 illustrates how LFCt was measured in this study. The y-axis in the top panel of Figure 2 is a binary variable indicating the presence or absence of respiratory events aligned at the event termination. The LFCt was calculated by aligning the SpO2 profile for each event at event termination and then ensemble averaging. LFCt is the time between the minimum of the ensemble-averaged SpO2 and event termination (time 0).4 This aggregate average LFCt was derived per participant and was used for analysis.
Figure 1.
Conceptual illustration of the LFCt measurement based on a single apnea event. LFCt = lung to finger circulation time; SpO2 = oxygen saturation as measured by pulse oximetry.
Figure 2.
Illustration of the ensemble-averaged LFCt. All respiratory events (and the associated SpO2 signals) were time-locked at their termination (time 0). The solid line in the top panel shows the probability of a respiratory event at any time point (the average of absence or presence of respiratory event at any time). The solid line in the bottom panel shows the average of associated SpO2 signal. The arrow shows the LFCt, the interval between the end of the respiratory events (time 0) and the minimum value of ensemble-averaged SpO2 signal (time 10 s in this example). The reason that the lines go up and down before time 0 is the differences between event durations for different events. Once all events and their associated SpO2 profiles were aligned at event termination, an ensemble averaging was performed, and the solid lines were obtained. The time between end of events (time 0) and minimum saturation reflects the LFCt. LFCt = lung to finger circulation time; SpO2 = oxygen saturation as measured by pulse oximetry.
Statistical Analysis
The primary outcomes were all-cause and CV mortality. Cause of death was broadly categorized by International Classification of Diseases, Ninth Revision codes as CV (codes 396.9-442, 966.71, 785.51), cancer (codes 141.9-208.0), and other causes (reported codes not in previous categories). Participants were contacted every 4 months, and more than 97% of contacts were completed. Next of kin were contacted for nonresponders. All reported deaths were centrally confirmed with death certificates and with medical records when available. The average ± SD follow-up was 9.9 ± 3.5 years. Cause-specific mortality, including deaths from CVD, was adjudicated by physicians at the coordinating center.
Covariates including demographic characteristics (race, age), anthropometric characteristics (BMI), health habits (smoking, alcohol use), and comorbid conditions (diabetes, hypertension, CVD, chronic obstructive lung disease, chronic kidney disease) were obtained by means of self-report of physician diagnosis. Prevalent CVD was defined as having a history of coronary heart disease (including myocardial infarction, coronary artery bypass grafting, percutaneous coronary intervention, or angina), HF, or stroke. History of physician-diagnosed CVD was obtained via questionnaire at the time of the sleep visit. Hypertension was defined according to those who self-reported hypertension (by affirmative answer to the question “Have you ever had hypertension?”), use of antihypertensive medications, or measured systolic BP ≥ 140 mm Hg or diastolic BP ≥ 90 mm Hg during the study examination. All analyses were also adjusted for study sites.
Baseline characteristics were compared according to LFCt quartiles. LFCt quartiles were the primary independent variables, with the lowest quartile being used as a reference group. Unadjusted cumulative incidence functions per quartile of LFCt were created for all-cause and CVD mortality. Cox proportional hazards regression models were constructed to calculate the hazard ratios (HRs) for the primary outcomes, adjusting for the aforementioned covariates without or with AHI (models 1 and 2, respectively). The motivation for adding AHI in model 2 was to examine whether the association might be influenced by AHI, the most commonly used metric to assess SDB. Because nocturnal hypoxemia is another important marker of SDB severity, in a separate model (model 3), we additionally adjusted for T90.
Using model 1 as a base model, we repeated analyses after excluding those with CVD at baseline without or with AHI (models 4 and 5, respectively), a restricted cohort. The rationale for this method was to examine the association in patients without known preexisting CVD, particularly with history of HF. People with HF often exhibit CSA-CSR, which inherently features long Ct. Furthermore, given a previous study showing an association of β-blocker use with long Ct, we performed a sensitivity analysis additionally adjusting for β-blocker and CSR.8 In addition, because hypoxic burden was a novel predictor of mortality and incident HF in recent studies, we performed an additional sensitivity analysis including hypoxic burden in the model.4,11 For these sensitivity analyses, LFCt was modeled as a continuous variable. Furthermore, regardless of the presence or absence of statistically significant interaction, to explore the impetus of LFCt on the basis of the SDB severity status, we examined the association of LFCt with outcomes stratified by presence of clinically significant (vs nonsignificant) SDB (AHI ≥ 15 vs < 15/h). Multicollinearity was absent, and the assumptions of proportionality were met. Finally, the type III P values for CVD and all-cause mortality models including all covariates, AHI, and continuous LFCt were calculated. The R statistical package (http://www.r-project.org) was used; the “car” package and analysis of variance function were used to calculate a type III P value. P < .05 was considered statistically significant.
Results
After we excluded participants without available death status and valid LFCt measurement, a total of 2,631 men (mean ± SD age, 76.4 ± 5.5 years) were available for analysis (Fig 3). Covariates were incomplete in 19 participants, and reported AHI was not available in 32 participants, resulting in a lower number for certain analysis models. Overall, reflecting the older age of the cohort and criteria for inclusion in this analysis, SDB was highly common, with a median (interquartile range) AHI of 18 (10-30) events per hour. Predominant CSA with CSR was rare (6.8%). LFCt was normally distributed, with a mean value of 19.0 s and median (interquartile range) of 18 (15-22) s. Men with high LFCt tended to be more obese and had a higher prevalence of hypertension, coronary heart disease, and HF than did those with lower LFCt (Table 1).
Figure 3.
Participant inclusion flowchart. MrOS = Osteoporotic Fractures in Men; PSG = polysomnography.
Table 1.
Baseline Characteristics
| Characteristic | All |
Lung to Finger Circulation Time Quartiles, s |
P Value | |||
|---|---|---|---|---|---|---|
| 1 (4-15) |
2 (15-18) |
3 (18-22) |
4 (22-52) |
|||
| N = 2,631 | n = 706 | n = 673 | n = 681 | n = 571 | ||
| Age, mean (SD), y | 76.4 (5.5) | 75.0 (4.9) | 75.7 (5.2) | 76.7 (5.4) | 78.5 (5.9) | < .001 |
| Race, No. (%) | .007 | |||||
| White | 2,390 (90.8) | 625 (88.5) | 600 (89.2) | 636 (93.4) | 529 (92.6) | … |
| Black | 82 (3.1) | 28 (4.0) | 26 (3.9) | 19 (2.8) | 9 (1.6) | … |
| Other | 159 (6.0) | 53 (7.5) | 47 (7.0) | 26 (3.8) | 33 (5.8) | … |
| BMI, mean (SD), kg/m2 | 27.2 (3.8) | 26.6 (3.6) | 27.1 (3.6) | 27.5 (3.9) | 27.8 (4.1) | < .001 |
| Smoking, No. (%) | .11 | |||||
| Current | 45 (1.7) | 13 (1.8) | 19 (2.8) | 8 (1.2) | 5 (0.9) | … |
| Never | 1,058 (40.2) | 295 (41.8) | 272 (40.4) | 264 (38.8) | 227 (39.8) | … |
| Former | 1,527 (58.0) | 398 (56.4) | 382 (56.8) | 409 (60.0) | 339 (59.4) | … |
| Diabetes, No. (%) | 357 (13.6) | 88 (12.5) | 97 (14.4) | 88 (13.0) | 84 (14.7) | .57 |
| Hypertension, No. (%) | 1,311 (49.8) | 326 (46.2) | 307 (45.6) | 356 (52.3) | 322 (56.4) | < .001 |
| CHD, No. (%) | 775 (29.5) | 184 (26.1) | 178 (26.4) | 198 (29.1) | 215 (37.7) | < .001 |
| HF, No. (%) | 167 (6.3) | 27 (3.8) | 38 (5.6) | 47 (6.9) | 55 (9.6) | < .001 |
| Stroke, No. (%) | 98 (3.7) | 20 (2.8) | 22 (3.3) | 26 (3.8) | 30 (5.3) | .13 |
| COPD, No. (%) | 136 (5.2) | 46 (6.5) | 31 (4.6) | 32 (4.7) | 27 (4.7) | .31 |
| CKD, No. (%) | 26 (1.0) | 6 (0.9) | 8 (1.2) | 8 (1.2) | 4 (0.7) | .77 |
| Total sleep time, mean (SD), min | 356 (69.2) | 360 (69.4) | 361 (69.3) | 353 (66.5) | 347 (71.1) | < .001 |
| AHI, events/h, median IQR) | 17.2 (10.1-28.2) | 15.0 (8.9-24.7) | 17.2 (10.7-27.7) | 18.5 (10.8-31.9) | 18.4 (10.2-29.8) | < .001 |
| T90, mean (SD), min | 4.32 (9.63) | 3.58 (8.91) | 4.26 (10.20) | 5.24 (10.30) | 4.22 (8.79) | .02 |
| T85, mean (SD), min | 0.48 (1.75) | 0.35 (1.46) | 0.49 (1.96) | 0.56 (1.87) | 0.51 (1.63) | .14 |
| T80, mean (SD), min | 0.09 (0.52) | 0.07 (0.47) | 0.09 (0.54) | 0.11 (0.59) | 0.10 (0.47) | .57 |
Percentages may not total 100% because of rounding. AHI = apnea-hypopnea index; CHD = coronary heart disease; CKD = chronic kidney disease; HF = heart failure; IQR = interquartile range; SpO2 = oxygen saturation as measured by pulse oximetry; T80 = time spent with SpO2 < 80%; T85 = time spent with SpO2 < 85%; T90 = time spent with SpO2 < 90%.
During an average ± SD follow-up of 9.9 ± 3.5 years, 427 (16%) and 1,205 (46%) of the sample experienced CV death and all-cause death, respectively. In quartile-based unadjusted analysis, men in the fourth quartile had a higher risk of both CV and all-cause mortality compared with men in the other three quartiles (Fig 4). In multivariate analysis, men in the fourth quartile had an increased risk of both CV mortality and all-cause mortality in partially and fully adjusted models (Figs 5, 6). Men in the fourth quartile of LFCt (22-52 s) had an HR of 1.36 (95% CI, 1.02-1.81) and 1.35 (95% CI, 1.14-1.60) for CV and all-cause mortality, respectively, when compared with men in the first quartile (4-15 s) (Table 2). The results were similar when additionally adjusting for AHI (model 2) (Figs 5, 6, Table 2) or for both AHI and T90 (model 3) (Table 2). In model 2, including AHI, no associations were seen between AHI and either CV or all-cause mortality. In model 3, including AHI and T90, no associations were seen between T90 and CV mortality, but the association existed, albeit weak, with all-cause mortality.
Figure 4.
Cumulative probability of mortality per quartile of LFCt. A, Cardiovascular disease mortality. B, All-cause mortality. LFCt = lung to finger circulation time; Q = quartile.
Figure 5.
CVD mortality hazard ratios based on lung to finger circulation time quartiles according to different covariate models. AHI = apnea-hypopnea index; CVD = cardiovascular; HTN = hypertension; T90 = time spent with oxygen saturation as measured by pulse oximetry < 90%.
Figure 6.
All-cause mortality hazard ratios based on lung to finger circulation time quartiles according to different covariate models. AHI = apnea-hypopnea index; CVD = cardiovascular; HTN = hypertension; T90 = time spent with oxygen saturation as measured by pulse oximetry < 90%.
Table 2.
Multivariate Analysis of All Men
| Model and Mortality | LFCt Quartile | HR (95% CI) | P Value |
|---|---|---|---|
| Model 1 (n = 2,612) | |||
| CV mortality | 1 | Reference | … |
| 2 | 1.02 (0.76-1.37) | .89 | |
| 3 | 0.92 (0.68-1.23) | .56 | |
| 4 | 1.36 (1.02-1.81) | .04 | |
| All-cause mortality | |||
| 1 | Reference | … | |
| 2 | 1.02 (0.87-1.21) | .78 | |
| 3 | 0.98 (0.83-1.17) | .86 | |
| 4 | 1.35 (1.14-1.60) | < .001 | |
| Model 2 (n = 2,580) | |||
| CV mortality | 1 | Reference | … |
| 2 | 1.00 (0.75-1.35) | .97 | |
| 3 | 0.89 (0.67-1.20) | .46 | |
| 4 | 1.36 (1.02-1.81) | .04 | |
| All-cause mortality | |||
| 1 | Reference | … | |
| 2 | 1.02 (0.86-1.21) | .82 | |
| 3 | 0.97 (0.82-1.14) | .69 | |
| 4 | 1.34 (1.13-1.59) | < .001 | |
| Model 3 (n = 2,580) | |||
| CV mortality | 1 | Reference | … |
| 2 | 1.00 (0.75-1.35) | .98 | |
| 3 | 0.89 (0.66-1.20) | .44 | |
| 4 | 1.36 (1.02-1.81) | .04 | |
| All-cause mortality | |||
| 1 | Reference | … | |
| 2 | 1.02 (0.86-1.21) | .85 | |
| 3 | 0.96 (0.81-1.14) | .66 | |
| 4 | 1.35 (1.14-1.60) | < .001 |
Cox proportional hazards regression models were used. Boldface indicates significant findings.
Model 1 adjusts for study site, race, age, BMI, smoking, alcohol use, diabetes, hypertension, coronary heart disease, heart failure, stroke, chronic obstructive lung disease, and chronic kidney disease.
Model 2 adjusts for model 1 covariates and apnea-hypopnea index.
Model 3 adjusts for model 2 covariates and time spent with oxygen saturation as measured by pulse oximetry < 90%.
CV = cardiovascular; HR = hazard ratio; LFCt = lung to finger circulation time.
After excluding men with baseline CVD (restricted cohort), the LFCt associations with mortality became stronger: Men in the fourth quartile had HRs of 1.74 (95% CI, 1.13-2.68) and 1.40 (95% CI, 1.12-1.75) for CV and all-cause mortality, respectively (model 4) (Table 3). Additional adjustment with AHI in this restricted cohort (model 5) yielded similar results (Table 3). In the stratified analysis of those with and those without SDB (AHI ≥ 15/h vs < 15/h), an association of LFCt with CV mortality was present only in those with SDB, whereas an association with all-cause mortality was present regardless of the SDB status (Table 4). Sensitivity analysis adjusting for β-blocker use or CSR yielded similar results (data not shown). In additional sensitivity analysis that further adjusted for the hypoxic burden in model 1, LFCt remained significant (CVD mortality HR, 1.13 [95% CI, 1.03-1.25]; all-cause mortality HR, 1.12 [95% CI, 1.06-1.19] per 1 SD LFCt). In these models that included both LFCt and hypoxic burden, hypoxic burden remained a significant predictor of CVD and all-cause mortality. The type III P values for LFCt effects on CVD and all-cause mortality were .004 and < .0001, respectively.
Table 3.
Multivariate Analysis Excluding Men With Prevalent CV Disease at Baseline
| Model and Mortality | LFCt Quartile | HR (95% CI) | P Value |
|---|---|---|---|
| Model 4 (n = 1,727) | |||
| CV mortality | 1 | Reference | … |
| 2 | 1.41 (0.92-2.17) | .11 | |
| 3 | 1.24 (0.80-1.90) | .34 | |
| 4 | 1.74 (1.13-2.68) | .01 | |
| All-cause mortality | |||
| 1 | Reference | … | |
| 2 | 0.99 (0.80-1.24) | .95 | |
| 3 | 1.11 (0.89-1.38) | .35 | |
| 4 | 1.40 (1.12-1.75) | .003 | |
| Model 5 (n = 1,706) | |||
| CV mortality | 1 | Reference | … |
| 2 | 1.40 (0.92-2.15) | .12 | |
| 3 | 1.21 (0.78-1.87) | .39 | |
| 4 | 1.76 (1.14-2.72) | .01 | |
| All-cause mortality | |||
| 1 | Reference | … | |
| 2 | 1.00 (0.81-1.25) | .97 | |
| 3 | 1.10 (0.88-1.36) | .40 | |
| 4 | 1.41 (1.13-1.77) | .002 |
Cox proportional hazards regression models were used. Boldface indicates significant findings.
Model 4 adjusts for model 1 covariates in a restricted cohort (excluding those with prevalent CV disease at baseline).
Model 5 adjusts for model 4 covariates and apnea-hypopnea index.
Model 1 adjusts for study site, race, age, BMI, smoking, alcohol use, diabetes, hypertension, coronary heart disease, heart failure, stroke, chronic obstructive lung disease, and chronic kidney disease.
CV = cardiovascular; HR = hazard ratio; LFCt = lung to finger circulation time.
Table 4.
LFCt and Mortality by Sleep Apnea Status
| Mortality | LFCt | HR (95% CI) | P Value |
|---|---|---|---|
| CV mortality | AHI < 15/h (n = 1,103) | ||
| Quartile 1 (4-15 s) | Reference | … | |
| Quartile 2 (15-18 s) | 0.76 (0.47-1.24) | .28 | |
| Quartile 3 (18-22 s) | 0.89 (0.57-1.40) | .63 | |
| Quartile 4 (22-52 s) | 1.10 (0.70-1.72) | .68 | |
| AHI ≥ 15/h (n = 1,477) | |||
| Quartile 1 (4-15 s) | Reference | … | |
| Quartile 2 (15-18 s) | 1.19 (0.80-1.78) | .38 | |
| Quartile 3 (18-22 s) | 0.93 (0.62-1.40) | .73 | |
| Quartile 4 (22-52 s) | 1.59 (1.07-2.37) | .02 | |
| All-cause mortality | AHI < 15/h (n = 1,103) | ||
| Quartile 1 (4-15 s) | Reference | … | |
| Quartile 2 (15-18 s) | 0.94 (0.73-1.23) | .69 | |
| Quartile 3 (18-22 s) | 1.02 (0.79-1.32) | .87 | |
| Quartile 4 (22-52 s) | 1.38 (1.07-1.79) | .01 | |
| AHI ≥ 15/h (n = 1,477) | |||
| Quartile 1 (4-15 s) | Reference | … | |
| Quartile 2 (15-18 s) | 1.04 (0.83-1.31) | .72 | |
| Quartile 3 (18-22 s) | 0.92 (0.73-1.16) | .47 | |
| Quartile 4 ([22-52 s) | 1.32 (1.05-1.66) | .02 | |
A Cox proportional hazards regression model was used for model 1, which adjusts for study site, race, age, BMI, smoking, alcohol use, diabetes, hypertension, coronary heart disease, heart failure, stroke, chronic obstructive lung disease, and chronic kidney disease, in all participants (n = 2,580). Boldface indicates significant findings. AHI = apnea-hypopnea index; CV = cardiovascular; HR = hazard ratio; LFCt = lung to finger circulation time.
Discussion
LFCt derived from routine sleep studies predicts all-cause and CVD mortality independent of other competing factors, including baseline CVD in elderly men, the vast majority of whom had at least mild SDB. Older men with an average LFCt > 22 s (the highest quartile) had at least a 35% increased risk of both CV and all-cause mortality compared with those with an LFCt < 15 s (the lowest quartile). The results were more pronounced after excluding men with prevalent CVD and were independent of conventional metrics of SDB and OSA-specific hypoxic burden. To our knowledge, this is the first large-scale community-based study to demonstrate the potential prognostic value of sleep study-derived LFCt.
A concept of Ct was established nearly a century ago and had been used for cardiac assessment before more advanced diagnostic technologies emerged.13 Ct was originally conceptualized as representing the mean velocity of flow of an injected substance (indicator) from the point of injection to the place of detection, enabling estimation of flow (“circulation”).14,15 Although direct use of Ct in clinical medicine is rare these days, the concept remains valid and forms a basis for other indicator-dilution-based cardiac output measurements such as thermodilution techniques and for timing of image acquisition in contrast-material-based modern imaging modalities.16, 17, 18 With use of oxygen as an indicator, the lung as an injection site, and the periphery as a detection site, Ct can be estimated and used to estimate cardiac output.19, 20, 21 In a similar manner, lung to periphery time can be obtained from a sleep study in the presence of clearly defined apneic or hypopneic episodes with accompanying oxygen desaturation.8 Prolonged Ct or delayed Ct, measured from a sleep study, is considered to reflect a low cardiac output state and is an important physiologic marker of CSA-CSR in patients with HF. Prior clinic-based studies of patients with OSA showed that Ct is more prolonged in patients with HF than in those without.8,22 Therefore, it is possible that older men with prolonged Ct may be a group with evidence of subclinical or undiagnosed HF.
LFCt predicted mortality consistently in different analytical models. The association was independent of AHI or T90. Furthermore, the association was independent of hypoxic burden, which was previously shown to be associated with mortality in this cohort.4 However, the association of hypoxic burden with mortality remained independent of LFCt, indicating their independent associations with mortality in this cohort. We found that the association of LFCt with mortality was more pronounced when participants with history of CVD were excluded. This finding may be due to attenuating effect of competing risk factors (CVD) on mortality. When stratified analyses were performed based on the presence of clinically significant SDB, LFCt was associated with all-cause mortality regardless of the SDB status. However, the association with CV mortality was seen only in the group with AHI ≥ 15/h. This finding may be due to lack of power but may also suggest that there may be synergistic effect in the presence of SDB in terms of CV mortality. As known, OSA, the most common phenotype of SDB, is characterized by recurrent obstruction of the airway, resulting in airflow restriction and typically accompanying intermittent hypoxemia. In addition, obstructive breathing elicits exaggerated intrathoracic pressure changes and a strong sympathetic response that culminates in a reduction in cardiac output through instantaneously increased afterload and systemic vascular resistance.23, 24, 25 In this regard, sleep study-derived LFCt is a novel measurement that may be useful in risk stratification of SDB independent of conventional metrics.
SDB is an important independent CV risk factor. In patients with existing CVD, SDB confers adverse outcomes.26,27 However, a randomized controlled trial was unable to detect a benefit of CPAP therapy in secondary prevention in patients with CVD.28 Similarly, in patients with HF and CSA, suppression of CSA with adaptive servo ventilation did not improve outcomes in this population.28 One caveat is that the AHI used in these studies may not be the best metric to classify patients for treatment allocation. Therefore, LFCt may be a useful marker to identify the highest future CV risk group who may benefit most from therapy. Secondary data analysis using existing cohorts or clinical trial data can provide more insights into this possibility. In this regard, a recent study in the Multi-Ethnic Study of Atherosclerosis cohort showing modest improvement in risk classification of coronary artery calcification when average LFCt is added to AHI as compared with AHI alone is noteworthy.29 In terms of the potential of LFCt as a marker of cardiac function, a recent work revealed that treatment of acute decompensated HF was associated with shortened LFCt, supporting the notion that LFCt is partially a function of cardiac function.30 Whether treatment of SDB would affect Ct over time in patients with or without HF is unknown and should be an area of future investigation. In addition, the mechanism by which LFCt is associated with the increased risk for all-cause mortality, beyond CV mortality, is unclear. However, it is possible that a substantial portion of non-CV mortality may have been CV mortality via subclinical CV impairment. However, the implication of abnormal LFCt in relation to non-CV mortality warrants further investigation.
In our quartile-based analysis, there was no consistent linear trend across the quartiles. The association of LFCt with outcomes was absent in either the second or third quartiles of LFCt. Rather, the association became evident in the fourth quartile, suggesting that there may be a threshold effect wherein a certain level of LFCt prolongation is necessary to be useful as a predictor of mortality. In the analysis excluding those with prevalent CVD at baseline, we found that older men in the fourth quartile (LFCt > 22 s) had approximately a 75% increased risk for CV mortality and a 41% increased risk for all-cause mortality. However, the thresholds used in this sample may not generalize to other samples, and future work can further characterize the distribution of LFCt across more diverse samples and identify the best ways to adjust for demographic characteristics and appropriate threshold measures. For LFCt to be a useful marker in the clinical setting, studies examining the distribution of absolute LFCt values in various populations are needed. The overall mean LFCt from our study was comparable with other studies.8,31 In the clinic-based study of young middle-aged adults (mean age, 48 years), the mean LFCt was 18.5 s, whereas in the Multi-Ethnic Study of Atherosclerosis consisting of older adults (median age, 69 years), the mean LFCt was 19.0 s. In both studies, older age and being male were associated with higher LFCt values.8,31 The LFCt value used in this study is an average from multiple measurements per subject. Therefore, LFCt from a single random event or limited number of events may have yielded different results. Furthermore, there is the possibility of change in the average LFCt over time even in the same individual because of change in volume status. 14,32
Finally, LFCt can be affected by the averaging time or smoothing techniques used by a particular pulse oximeter. Although not relevant in this study because same pulse oximeter was used for all the PSG studies, some degree of error in measurement should be considered when comparing LFCt values between studies using different models. In addition, the reproducibility of LFCt, such as internight variability, needs to be verified to be a valid marker. Despite these limitations and gaps, given the readily available nature of LFCt as a metric, we propose that the PSG-derived LFCt may be a novel prognostic marker and may be a useful tool that can guide clinical decision-making.
Our study has several strengths, including the large size of the community-based population of older men and objective measurements of SDB by means of full PSG. Use of automated LFCt with ensemble averaging allowed consistent measurement, unlike manual measurement that is susceptible to interrater inconsistency. This analysis also involved a long follow-up with a high number of events, allowing for sufficient power. Several limitations are worthy of mention. Information about baseline CVD, including HF, was based on self-report of physician diagnosis. The results are based on a single night’s study, so the test-retest reliability of LFCt cannot be ascertained in this study. Finally, this study included only elderly men, so the results may not be generalizable to women, younger men, and more diverse ethnicity groups.
Conclusions
In conclusion, our prospective analysis showed that sleep study-derived Ct (LFCt) was independently associated with both CV and all-cause mortality in older men with SDB, independent of both baseline CV burden and conventional SDB metrics. LFCt may be a novel physiologic marker for CV vulnerability and adverse outcomes in patients with SDB.
Acknowledgments
Author contributions: Conceptual design: Y. K., A. W., S. R., and A. A.; Analysis: Y. K. and A. A.; Critical review and manuscript writing: Y. K., S. S., K. S., L. T., R. A., D. W., A. W., S. R. and A. A.
Financial/nonfinancial disclosures: The authors have reported to CHEST the following: S. A. S. receives personal fees as a consultant for Merck & Co and Nox Medical outside the submitted work and receives grant support from Apnimed and ProSomnus Sleep Technologies. K. L. S. has grant funding from Merck & Co. L. T. M. works as a consultant for Apnimed and has a financial interest in Apnimed, a company developing pharmacologic therapies for sleep apnea; his interests were reviewed and are managed by Brigham and Women’s Hospital and Partners HealthCare in accordance with their conflict of interest policies. D. P. W. is the chief scientific officer for Philips Respironics. A. W. works as a consultant for Apnimed, Nox Medical, and SomniFix; has received grants from Sanofi and SomniFix; and has a financial interest in Apnimed, a company developing pharmacologic therapies for sleep apnea; his interests were reviewed and are managed by Brigham and Women’s Hospital and Partners HealthCare in accordance with their conflict of interest policies. S. R. received grant support and consulting fees from Jazz Pharmaceuticals plc and a consulting fee from Eisai Co, Ltd, and Respircardia. A. A. serves as consultant for Apnimed and SomniFix and reports a grant from SomniFix. None declared (Y. K., R. M. A.).
Role of sponsors: The sponsor had no role in the design of the study, the collection and analysis of the data, or the preparation of the manuscript.
Footnotes
FUNDING/SUPPORT: The Osteoporotic Fractures in Men (MrOS) study is supported by the National Institutes of Health (NIH) through the National Institute of Arthritis and Musculoskeletal and Skin Diseases, National Institute on Aging, National Center for Research Resources, and NIH Roadmap for Medical Research [Grants U01 AR45580, U01 AR45614, U01 AR45632, U01 AR45647, U01 AR45654, U01 AR45583, U01 AG18197, U01-AG027810, UL1 TR000128, K24-AR04884-06]. The National Heart, Lung, and Blood Institute (NHLBI) provides funding for the MrOS Sleep ancillary study, Outcomes of Sleep Disorders in Older Men [Grants R01 HL071194, R01 HL070848, R01 HL070847, R01 HL070842, R01 HL070841, R01 HL070837, R01 HL070838, R01 HL070839]. Y. K. was partly supported by the NHLBI [Grant R21 HL140432]. S. A. S., A. W., S. R., and A. A. were partially supported by the NHLBI [Grant R35HL135818]. A. A. was supported by the American Heart Association [Grant 19CDA34660137], NHLBI [Grant 1R01HL153874], and American Academy of Sleep Medicine Foundation [Grant 188-SR-17].
References
- 1.Gottlieb D.J., Yenokyan G., Newman A.B. A prospective study of obstructive sleep apnea and incident coronary heart disease and heart failure: the Sleep Heart Health Study. Circulation. 2010;122(4):352–360. doi: 10.1161/CIRCULATIONAHA.109.901801. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Peppard P.E., Young T., Palta M., Skatrud J. Prospective study of the association between sleep-disordered breathing and hypertension. N Engl J Med. 2000;342(19):1378–1384. doi: 10.1056/NEJM200005113421901. [DOI] [PubMed] [Google Scholar]
- 3.Redline S., Yenokyan G., Gottlieb D.J. Obstructive sleep apnea-hypopnea and incident stroke: the Sleep Heart Health Study. Am J Respir Crit Care Med. 2010;182(2):269–277. doi: 10.1164/rccm.200911-1746OC. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Azarbarzin A., Sands S.A., Stone K.L. The hypoxic burden of sleep apnoea predicts cardiovascular disease-related mortality: the Osteoporotic Fractures in Men Study and the Sleep Heart Health Study. Eur Heart J. 2019;40(14):1149–1157. doi: 10.1093/eurheartj/ehy624. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Jonas D.E., Amick H.R., Feltner C. Screening for obstructive sleep apnea in adults: evidence report and systematic review for the US Preventive Services Task Force. JAMA. 2017;317(4):415–433. doi: 10.1001/jama.2016.19635. [DOI] [PubMed] [Google Scholar]
- 6.Punjabi N.M. Counterpoint: is the apnea-hypopnea index the best way to quantify the severity of sleep-disordered breathing? No. Chest. 2016;149(1):16–19. doi: 10.1378/chest.14-2261. [DOI] [PubMed] [Google Scholar]
- 7.Hall M.J., Xie A., Rutherford R., Ando S., Floras J.S., Bradley T.D. Cycle length of periodic breathing in patients with and without heart failure. Am J Respir Crit Care Med. 1996;154(2 pt 1):376–381. doi: 10.1164/ajrccm.154.2.8756809. [DOI] [PubMed] [Google Scholar]
- 8.Kwon Y., Khan T., Pritzker M., Iber C. Circulation time measurement from sleep studies in patients with obstructive sleep apnea. J Clin Sleep Med. 2014;10(7):759–765. doi: 10.5664/jcsm.3866. 765A. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Orwoll E., Blank J.B., Barrett-Connor E. Design and baseline characteristics of the Osteoporotic Fractures in Men (MrOS) study: a large observational study of the determinants of fracture in older men. Contemp Clin Trials. 2005;26(5):569–585. doi: 10.1016/j.cct.2005.05.006. [DOI] [PubMed] [Google Scholar]
- 10.Blank J.B., Cawthon P.M., Carrion-Petersen M.L. Overview of recruitment for the Osteoporotic Fractures in Men study (MrOS) Contemp Clin Trials. 2005;26(5):557–568. doi: 10.1016/j.cct.2005.05.005. [DOI] [PubMed] [Google Scholar]
- 11.Azarbarzin A., Sands S.A., Taranto-Montemurro L. The sleep apnea-specific hypoxic burden predicts incident heart failure. Chest. 2020;158(2):739–750. doi: 10.1016/j.chest.2020.03.053. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Redline S., Sanders M.H., Lind B.K. Methods for obtaining and analyzing unattended polysomnography data for a multicenter study: Sleep Heart Health Research Group. Sleep. 1998;21(7):759–767. [PubMed] [Google Scholar]
- 13.Diettert G.A. Circulation time in the aged. JAMA. 1963;183(12):1037–1038. doi: 10.1001/jama.1963.63700120034017. [DOI] [PubMed] [Google Scholar]
- 14.Pearce M.L., Lewis A.E., Kaplan M.R. The factors influencing the circulation time. Circulation. 1952;5(4):583–588. doi: 10.1161/01.cir.5.4.583. [DOI] [PubMed] [Google Scholar]
- 15.Blumgart H.L., Weiss S. Clinical studies on the velocity of blood flow: IX—the pulmonary circulation time, the velocity of venous blood flow to the heart, and related aspects of the circulation in patients with cardiovascular disease. J Clin Invest. 1928;5(2):343–377. doi: 10.1172/JCI100164. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Fox I.J. History and developmental aspects of the indicator-dilution technic. Circ Res. 1962;10:381–392. doi: 10.1161/01.res.10.3.381. [DOI] [PubMed] [Google Scholar]
- 17.Bae K.T. Intravenous contrast medium administration and scan timing at CT: considerations and approaches. Radiology. 2010;256(1):32–61. doi: 10.1148/radiol.10090908. [DOI] [PubMed] [Google Scholar]
- 18.Zierler K. Indicator dilution methods for measuring blood flow, volume, and other properties of biological systems: a brief history and memoir. Ann Biomed Eng. 2000;28(8):836–848. doi: 10.1114/1.1308496. [DOI] [PubMed] [Google Scholar]
- 19.Kasravi B., Boehmer J.P., Leuenberger U.A. A noninvasive method for estimating cardiac output using lung to finger circulation time of oxygen. Am J Cardiol. 1998;82(7):915–917. doi: 10.1016/s0002-9149(98)00505-0. [DOI] [PubMed] [Google Scholar]
- 20.Kwon Y., Van’t Hof J., Roy S.S., Bache R.J., Das G. A novel method for assessing cardiac output with the use of oxygen circulation time. J Card Fail. 2016;22(11):921–924. doi: 10.1016/j.cardfail.2015.10.021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Wexler J., Whittenberger J.L., Himmelfarb S. An objective method for determining circulation time from pulmonary to systemic capillaries by the use of the oximeter. J Clin Invest. 1946;25(3):447–450. doi: 10.1172/JCI101727. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Ryan C.M., Bradley T.D. Periodicity of obstructive sleep apnea in patients with and without heart failure. Chest. 2005;127(2):536–542. doi: 10.1378/chest.127.2.536. [DOI] [PubMed] [Google Scholar]
- 23.Garpestad E., Katayama H., Parker J.A. Stroke volume and cardiac output decrease at termination of obstructive apneas. J Appl Physiol (1985) 1992;73(5):1743–1748. doi: 10.1152/jappl.1992.73.5.1743. [DOI] [PubMed] [Google Scholar]
- 24.Somers V.K., Dyken M.E., Clary M.P., Abboud F.M. Sympathetic neural mechanisms in obstructive sleep apnea. J Clin Invest. 1995;96(4):1897–1904. doi: 10.1172/JCI118235. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Tolle F.A., Judy W.V., Yu P.L., Markand O.N. Reduced stroke volume related to pleural pressure in obstructive sleep apnea. J Appl Physiol Respir Environ Exerc Physiol. 1983;55(6):1718–1724. doi: 10.1152/jappl.1983.55.6.1718. [DOI] [PubMed] [Google Scholar]
- 26.Wilcox I., McNamara S.G., Wessendorf T., Willson G.N., Piper A.J., Sullivan C.E. Prognosis and sleep disordered breathing in heart failure. Thorax. 1998;53(suppl 3):S33–S36. doi: 10.1136/thx.53.2008.s33. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Zhao Y., Yu B.Y., Liu Y., Liu Y. Meta-analysis of the effect of obstructive sleep apnea on cardiovascular events after percutaneous coronary intervention. Am J Cardiol. 2017;120(6):1026–1030. doi: 10.1016/j.amjcard.2017.06.035. [DOI] [PubMed] [Google Scholar]
- 28.McEvoy R.D., Antic N.A., Heeley E. CPAP for prevention of cardiovascular events in obstructive sleep apnea. N Engl J Med. 2016;375(10):919–931. doi: 10.1056/NEJMoa1606599. [DOI] [PubMed] [Google Scholar]
- 29.Kwon Y., Mariani S., Reid M. Lung to finger circulation time in sleep study and coronary artery calcification: the Multi-Ethnic Study of Atherosclerosis. Sleep Med. 2020;75:8–11. doi: 10.1016/j.sleep.2020.05.033. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Tobushi T., Hosokawa K., Abe K. Changes in lung to finger circulation time measured via cardiopulmonary polygraphy in patients with varying types of heart disease. Heart Vessels. 2021;36(1):58–68. doi: 10.1007/s00380-020-01657-2. [DOI] [PubMed] [Google Scholar]
- 31.Kwon Y., Mariani S., Gadi S.R. Characterization of lung-to-finger circulation time in sleep study assessment: the Multi-Ethnic Study of Atherosclerosis. Physiol Meas. 2020;41(6) doi: 10.1088/1361-6579/ab8e12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Tarr L., Oppenheimer B.S., Sager R.V. The circulation time in various clinical conditions determined by the use of sodium dehydrocholate. Am Heart J. 1933;8(6):766–786. [Google Scholar]