The Influence of Intermittent Hypoxemia on Platelet Activation in Obese Patients with Obstructive Sleep Apnea

Study Design

Obese individuals were recruited from the community from October 2005 to September 2008. Following an outpatient screening visit to determine eligibility, subjects underwent a standard in-laboratory overnight polysomnogram. Subjects remained fasting from 22:00 until the study was completed the next morning and refrained from any vigorous physical activity. Blood collection for platelet testing was performed on the morning following the polysomnogram, approximately 2 hours after awakening. Flow cytometry samples were labeled for immunophenotypic analysis within 30 minutes of blood draw and were stabilized by fixation. Samples were analyzed on the flow cytometer within the next 24 to 48 hours, a time frame that does not result in artifactual platelet activation.10

Study Population

Obese individuals with a BMI ≥ 30 kg/m², aged 18 to 70 years, were enrolled if they were nonsmokers and free of cardiac, pulmonary, endocrine, or sleep disorders (other than hypertension in 3 patients and the presence of OSA). All subjects underwent thorough history and physical examination by a licensed physician and laboratory testing for fasting blood glucose concentration, fasting lipid panel, complete blood count, and thyroid stimulating hormone level. Subjects with abnormal results on physical exams or laboratory values were excluded. Prior to the study night, all subjects were asked to keep a 2-week sleep diary, and those with partial sleep deprivation or erratic sleep schedules were excluded. Subjects were also excluded if they were taking medications that could affect cardiovascular function or sleep, such as oral contraceptive agents, hormone replacement therapy, sedatives, steroids, antiplatelet drugs (including nonsteroidal antiinflammatory drugs), - antihypertensive medications, and lipid-lowering agents. Thus, we excluded any subjects with known preexisting or unstable vascular disease, inflammatory diseases, or infections that could affect platelet function. All subjects provided written informed consent. The study was approved by the Institutional Review Board and General Center for Clinical Research of Brigham and Women's Hospital.

Laboratory Measurements

Blood samples were obtained using standard sterile technique from the antecubital vein. Plasma glucose concentration, total serum cholesterol level, and triglyceride levels were measured using the Synchoron CX analyzer (Beckman Systems, Fullerton, CA). High-density lipoprotein serum cholesterol levels were measured directly (Sigma, St. Louis, MO), and low-density lipoprotein levels were calculated. The glycated hemoglobin (HbA_1c) concentration was measured in whole blood with ion-exchange high-performance liquid chromatography. Complete blood counts (CBC), thyroid stimulating hormone levels, and fasting glucose concentrations were measured in a core laboratory using standard laboratory techniques.

Polysomnography

Recorded polysomnographic signals included electroencephalogram (C4-A1, C3-A2, O2-A1, and O1-A2), left and right electrooculogram, submental and bilateral tibial electromyogram, electrocardiogram with surface electrodes, airflow, chest and abdominal excursion (piezo bands), oxyhemoglobin saturation (SaO₂), and body position. All data were collected and stored on Nihon Kohden, a digital polysomnography system (Nihon Kohden; Foothill Ranch, CA). Polysomnograms were scored by a blinded, experienced sleep technician and staged according to standard criteria.11 Arousals and respiratory events were scored according to published guidelines.12,13 An apnea was scored if airflow was absent for 10 seconds, and a hypopnea was scored if there was at least a 50% reduction in airflow for 10 seconds or a discernable decrement in airflow for 10 seconds in association with a either an oxyhemoglobin desaturation of at least 3% or an arousal. A respiratory-disturbance index (RDI) was then calculated based on the number of apneas and hypopneas per hour of sleep. The time spent with an SaO₂ below 90% was represented as minutes of desaturation time < 90% divided by the hours of total sleep time (TST) (T < 90%). The SaO₂ nadir represented the lowest oxygen saturation during sleep, and the oxygen desaturation index (ODI) was based on the number of desaturations (decrease of at least 4% from baseline) per hour of sleep. The TST, arousal index (AI), and sleep efficiency (SE) were determined after manual scoring of the polysomnographic record.

Measurement of Platelet Activation

Five biomarkers of platelet activation were measured by whole-blood flow cytometry in 3 states (no stimulation, stimulation with 5 μM ADP, and stimulation with 20 μM ADP): platelet-surface P-selectin, platelet surface-activated GPIIb/IIIa, platelet-surface GPIb, platelet-monocyte aggregates, and platelet-neutrophil aggregates. To minimize platelet activation during blood collection, blood samples were obtained with a 21-gauge needle into a sodium citrate blue-top tube. The initial 4 mL of blood drawn was not used for platelet testing, and the tube was handled with care to avoid inadvertent activation of platelets. Preparation of whole-blood samples for flow-cytometry analysis has been previously described in detail.14,15 In brief, samples for flow cytometry were prepared by a highly trained investigator within 30 minutes of blood's being drawn. For platelet-activation studies, whole-blood samples were labeled with phycoerythrin (PE)-labeled P-selectin-specific monoclonal antibody (CD62P, Santa Cruz Biotech, Santa Cruz, CA), fluorescein isothiocyanate (FITC)-labeled monoclonal antibody specific for activated GPIIb/IIIa (PAC-1, Becton Dickinson [BD], Franklin Lakes, NJ), PE-Cy5-labeled GPIb-specific monoclonal antibody (CD42b, BD-Pharmingen) and exposed to 1 of 3 activation conditions for 15 minutes: (1) buffer, (2) 0.5 μM ADP, (3) or 20 μM ADP. Following this exposure, platelet samples were fixed with HEPES-saline-1% ultrapure formaldehyde (Polysciences, Inc., Warrington, PA). Nonspecific PE-antibody binding was determined using mouse IgG1_2a-PE (Santa Cruz Biotech) in place of the P-selectin-specific antibody. Nonspecific FITC fluorescence was determined by adding 2.5 μg/mL eptifibatide (a small-molecule GPIIb/IIIa antagonist, Millennium Pharmaceuticals, Cambridge, MA) to block site-specific PAC-1–FITC binding. For platelet-monocyte and platelet-neutrophil aggregation studies, whole-blood samples were labeled with anti-CD14 (PE-Cy5, Beckman Coulter, Inc., Brea, CA) to identify monocytes (bright) and neutrophils (dim) and a PE-labeled GPIX-specific monoclonal antibody (CD42a, BD-Pharmingen) to identify bound platelets and then were exposed to 1 of 3 activation conditions for 15 minutes: (1) buffer, (2) 0.5 μM ADP, or (3) 20 μM ADP. Following this exposure, samples were fixed with Becton Dickinson FACS lysing solution, a single-step fix-lyse reagent. Nonspecific PE fluorescence was determined by using MIgG1κ-PE (BD-Pharmingen) in place of the anti-GPIX antibody. Samples were analyzed in a FACSCalibur flow cytometer (BD Biosciences, San Jose, CA) within 24 to 48 hours with 488-nm excitation and appropriate color-compensation settings. The flow cytometer was calibrated daily with Spherotech (Lake Forest, IL) calibration particles to adjust PMT voltages, which ensured consistent fluorescence values during the course of the study.

Data acquisition and analysis were performed with BD CellQuest Pro 2 Software (Becton Dickinson Biosciences). Platelets were identified by their characteristic light scatter and positive labeling with platelet-specific anti-GPIb monoclonal antibody,and these platelets were then assessed for percentage of platelets positive for PAC-1 binding and P-selectin and also assessed for mean fluorescence of PAC-1, P-selectin, and GPIb (indicating average receptor density). Monocytes and neutrophils were identified by their characteristic light scatter and CD14 fluorescence. Platelet-monocyte aggregates (PMA) and platelet-neutrophil aggregates (PNA) were defined by the percentage of monocytes or neutrophils that were positive for platelet-specific marker GPIX.

Statistical Analysis

Patient characteristics are represented as means and statistical differences and were calculated using t tests for 2-group comparison. Variables with significant departure from the normal distribution were presented as median and interquartile range and analyzed with nonparametric tests: Mann-Whitney for 2-group comparison and Kruskal-Wallis for comparison between larger groups. The Pearson χ² test was used for comparison among categorical variables. The biomarkers of platelet activation were assessed in relation to T < 90% status (dichotomized as explained below), ODI (dichotomized into groups above and below the median value of 4.5 events/h), OSA status, TST, AI, and SE. The T < 90% value was dichotomized based on a clinically significant time of 1 or more minutes per hour of sleep. A multiple linear-regression analysis was performed to determine if relevant OSA and sleep parameters in all subjects were associated with platelet activation, reactivity, and aggregability independently of the BMI, age, sex, and logAHI. To avoid model overfitting, we used forward stepwise regression to identify potential significant predictors (p < 0.10) for retention in the model. Only a portion of our subjects with OSA had significant desaturations; thus, in posthoc analysis, subjects were also separated into 3 groups: (1) those with no OSA, (2) those with OSA but minimal desaturation (T < 90% was < 1 min/h of sleep), and (3) those with OSA and desaturation (T < 90% was ≥ 1min/h of sleep). All reported p values are 2 sided, and p < 0.05 was considered to be significant.

Subject Characteristics

Of the 341 potential subjects who underwent screening, 77 met inclusion and exclusion criteria and were enrolled. Characteristics of the 77 subjects studied are shown in Table 1A. The groups were well matched for BMI, as intended, and had similar body-fat percentages. Differences in age and sex among the groups with and without OSA were statistically adjusted, as shown below. Based on an AHI of at least 10 events per hour to define OSA, 47 of the 77 studied subjects had OSA (see Table 1B). The median AHI in subjects with OSA was 24.7 ± 28.1 events/hour and 3.0 ± 3.9 events/hour in subjects without OSA (p < 0.001). Additionally, there were the expected differences between the 2 groups in the SaO₂ nadir, T < 90%, ODI, and AI. TST and SE were similar in both groups.

Platelet Activation

The relationships between platelet activation and OSA status, T < 90%, ODI, TST, AI, and SE were examined. Those subjects with desaturations above the cutoff of more than 1 minute of desaturation time showed a significant decrease in GPIb fluorescence intensity in circulating platelets and after mild and strong stimulation with ADP (with 0.5 μM and 20 μM of ADP) (Figure 1), indicating circulating platelet activation and increased platelet reactivity. Multiple linear regression analysis revealed that T < 90% and female sex were the only significant independent predictors of GPIb fluorescence intensity in a model that included T < 90% status, OSA status, age, sex, and BMI. Although OSA status (AHI ≥ 10 events/h) did not predict platelet activation, as determined by GPIb fluorescence intensity (Figure 2), most subjects with longer desaturation times also had more severe OSA (Figures 3A and 3B). However, the relationship between AHI and T < 90% was not linear. Thirteen subjects with OSA and desaturations (T < 90% was ≥ 1 min/h of sleep, Group 3) had more activated platelets than did the 29 subjects with OSA and minimal desaturations (T < 90% was < 1 min/h of sleep, Group 2) or the 27 subjects without OSA (Group 1, Figure 4). A greater ODI was not statistically significantly predictive of GPIb levels. There were no changes in other markers of platelet activation (platelet surface-activated GPIIb/IIIa, platelet surface P-selectin, PMA, and PNA) in relationship to OSA status, T < 90%, ODI, TST, AI, or SE.

Limitations

Although our study had a number of strengths, including our relatively large sample size for a physiological study, our rigorous exclusion of potential confounding variables and our use of gold-standard techniques (in-laboratory polysomnography and flow cytometry with response to agonists performed in a laboratory with considerable expertise in platelet biology), we acknowledge the following limitations. This is an observational study, and thus, associations of platelet function with various indices of OSA severity do not prove causality. Differences in platelet function may be related to unrecognized differences in subjects who have more severe desaturation, rather than to sleep apnea, per se. Additionally, samples were collected after the subject awakened and was in an upright position. Prior data suggest that the upright position activates platelets, and our study design, which included the collection of platelets after the subject assumed the upright position, could thus bias our findings toward the null hypothesis.4 However, this procedure was systematic (in both groups) and not altered throughout the study period. Additionally, in prior studies examining the role of OSA in platelet activation, blood was collected from patients shortly after they awakened, whereas we collected blood 2 hours after the subjects awakened. Biomarkers, such as platelet surface P-selectin and platelet surface-activated GPIIb/IIIa, remain activated for a shorter time and return to baseline values faster than does GPIb. Thus, timing of our blood draw (2 hours after awakening rather than immediately after awakening or during sleep) may also explain the lack of platelet activation in subjects with OSA, as assessed by some markers. We also accept that, because we used multiple metrics of platelet activation to provide a thorough assessment of platelet function, we performed multiple comparisons. Thus, any significant results may be purely due to chance alone, although we doubt that all of our findings are spurious. Finally, one could argue that our exclusion of covariates yielded an ‘exceptionally healthy’ sample of obese subjects, which could lead to a lack of generalizability of our results. For example, perhaps the concurrent presence of OSA, diabetes, and obesity all interactively affect platelet function. Although we accept that our conclusions are limited to the population studied, we believe that isolated variables should be assessed before these complex interactions can be defined in cohorts that would yield more generalizable results.

Clinical Significance

Our findings suggest that even relatively “healthy obese” subjects with OSA may have platelet activation if OSA is accompanied by desaturation. These findings suggest that, when assessing the severity of OSA and the potential for cardiovascular complications developing, desaturation time must be considered in addition to the AHI. The findings of the present study will require confirmation in other patient cohorts. Further studies may help stratify which patients with OSA should be targeted therapeutically with antiplatelet agents for the prevention of subsequent atherothrombotic disease.