Sleep-Disordered Breathing and Transcranial Dopplers in Sickle Cell Disease

Objectives To determine the prevalence of sleep-disordered breathing in children with sickle cell disease and whether there is an association of sleep-disordered breathing with high-risk transcranial Doppler ultrasonography (TCD) velocities.

Study Design Cross-sectional.

Setting Tertiary care academic medical center.

Patients Sixty-four children (aged 2-14 years) selected for eligible genotype (type SS or Sβ⁰-thalassemia) and no history of stroke.

Interventions Parents completed the Pediatric Sleep Questionnaire. Overnight polysomnography was performed for children with snoring. The TCD was performed or existing results were obtained for all children; for children who underwent transfusion therapy, readings prior to the transfusion were analyzed. Children with abnormal or conditional TCD (flow velocity ≥170 cm/s in any vessel) were considered high risk.

Main Outcome Measures Prevalence of sleep-disordered breathing and TCD velocity and frequency of high-risk TCD in patients with and without sleep-disordered breathing.

Results The prevalence of snoring was 37.5% (95% CI, 26.7%-49.8%), the prevalence of positive polysomnography findings was 23.7% (14.6%-36.1%), and the prevalence of positive Pediatric Sleep Questionnaire scores was 21.9% (13.4%-33.6%). There was no significant difference in TCD velocity or number of patients with high-risk TCD between nonsnorers and children with snoring but negative polysomnography findings and children with snoring and positive polysomnography findings (P = .91 and P = .66, respectively) or between nonsnorers and snorers with a negative Pediatric Sleep Questionnaire score and snorers with a positive Pediatric Sleep Questionnaire score (P = .76 and P = .33, respectively).

Conclusion There is a high prevalence of snoring and sleep-disordered breathing among children with sickle cell disease, but our results do not support an association with cerebrovascular risk.

Sickle cell disease (SCD) affects 1 in 350 African American newborns each year.¹ It is an inherited hemoglobinopathy in which deoxygenation leads to sickling of the red blood cells, causing vaso-occlusion with recurrent pain and chest episodes. In addition, vasculopathy develops, perhaps due to chronic hemolysis, resulting in chronic organ damage and specifically risk of stroke.²

Patients with SCD appear to be at increased risk for sleep-disordered breathing (SDB). One hypothesis suggests that this may be due to a compensatory hypertrophy of the tonsils and adenoids secondary to autosplenectomy.³ Sleep-disordered breathing is viewed as a continuum of severity, from partial obstruction of the airway producing snoring to increased upper airway resistance and finally episodes of complete upper airway obstruction or obstructive sleep apnea (OSA). Children with SCD and SDB may have a higher incidence of OSA, as well as more severe nocturnal desaturation and hypercapnia.⁴ The prevalence of SDB in the general pediatric population is between 1% and 3%, but the prevalence in patients with SCD has been estimated to be 10% to 36%.^5,6

The incidence of stroke in the population with SCD is approximately 11%.⁷ Transcranial Doppler ultrasonography (TCD) of the major intracranial vessels is a noninvasive screening method used to assess stroke risk in children with SCD; velocities greater than 170 cm/s are associated with progressively increased risk of stroke, which can be prevented by chronic transfusion therapy.^8,9 Oxygen desaturation and impaired dynamic cerebral autoregulation may predispose patients with SCD to stroke.^10,11 It has been proposed that OSA plays a pivotal role in precipitating nighttime hypoxemia and the vaso-occlusive episodes associated with SCD.¹² Sickle cell vaso-occlusion is caused by a complex interplay of hemoglobin S polymerization, red cell–endothelial cell interactions, hypercoagulability, neutrophil activation, and vasoactive factors. Molecular findings associated with OSA include endothelial dysfunction, increased C-reactive protein, interleukin-6, and increased clotting factors. These findings suggest that OSA may exacerbate sickle cell–related pathology, resulting in increased incidence or severity of both vaso-occlusion and vascular disease.

The purpose of this study was to establish the prevalence of SDB in children with SCD. We used the Pediatric Sleep Questionnaire (PSQ),¹³ a validated instrument used to identify SDB in children in clinical research settings, and overnight polysomnography (PSG), the criterion standard for diagnosis of SDB, to determine whether there is an association of snoring and SDB with increased TCD velocity. We also examined whether known interventions for SCD complications (chronic transfusion and hydroxyurea therapies) and known risk factors for SDB (age, sex, prematurity, asthma, body mass index, and socioeconomic status) are associated with SDB.

Children aged 2 to 14 with SCD (type SS or Sβ⁰-thalassemia) attending routine visits at the sickle cell clinic at the State University of New York Downstate Medical Center or the Kings County Hospital Center were offered enrollment in the study. Exclusion criteria were history of stroke (because TCDs are not performed), craniofacial syndromes, cerebral palsy, neuromuscular disease, mucopolysaccharide storage disease, immunodeficiency, mental or physical impairment severe enough to preclude interpretation of the behavioral information, or the inability of the parent, guardian, or caretaker to read or understand English. Children who had previously undergone treatment of SDB (adenotonsillectomy [AT], tonsillectomy, adenoidectomy, or nasal continuous positive airway pressure) were included because this was a point prevalence study, but data regarding treatment were collected. The study was approved by the institutional review boards of both centers. Informed consent was obtained from the parents or guardians, and assent was obtained from children older than 7 years. A convenience sample based on the availability of the researchers to attend clinic sessions was recruited.

Parents or guardians completed the Sleep-Related Breathing Disorder scale of the PSQ.¹³ The Sleep-Related Breathing Disorder scale consists of 22 closed-response question-items that ask about snoring frequency, loud snoring, observed apneas, difficulty breathing during sleep, daytime sleepiness, inattentive or hyperactive behavior, and other features of pediatric SDB. Each item is answered yes (1), no (0), or don't know (missing). The number of symptom-items endorsed positively (“yes”) is divided by the number of items answered positively or negatively; the denominator excludes items with missing responses and items answered as don't know. The result is a proportion that ranges from 0.0 to 1.0. Scores greater than 0.33 are considered positive and suggestive of high risk for pediatric SDB. The threshold is based on a validity study that suggested optimal sensitivity and specificity at the 0.33 cut-off.¹³

Children with snoring as determined by a positive response to question A2 (“Does your child snore more than half the time?”) were referred for overnight PSG testing because this item has the strongest association of all the questions regarding the frequency and severity of snoring with positive PSG.¹³ The PSG consisted of electroencephalography, electro-oculography, electromyography, electrocardiography, respiratory rate, pulse rate, and pulse oximetry; inductive plethysmography of the chest and abdomen and the mathematical sum of the two for respiratory effort; and oronasal airflow from nasal thermistors and nasal air pressure transducers. Obstructive apnea was defined as the cessation of oronasal airflow with continued respiratory effort for at least 2 times the typical breath interval, and obstructive hypopnea was defined as a decrease in amplitude of oronasal airflow of at least 50% with no decrease in respiratory effort for the same duration associated with an arousal, an awakening, or a greater than 3% desaturation. Polysomnography was considered positive for SDB if the apnea-hypopnea index was at least 2, the apnea index was at least 1, or the minimum oxygen saturation was less than 92%. The PSG was interpreted by a pulmonologist masked to TCD results. Children with positive PSG findings were referred to the otolaryngology clinic for consideration of AT.

All patient medical records were reviewed to determine whether TCD was performed in the past year, as recommended. Patients who did not have a TCD were referred for the procedure in the Division of Pediatric Hematology at State University of New York Downstate Medical Center. The highest time-averaged mean of the maximum flow velocity in 2-mm increments was measured in the middle cerebral artery, the distal internal carotid artery, the anterior and posterior cerebral arteries, and the basilar artery. Stroke Prevention Trial in Sickle Cell Anemia criteria were used to classify flow velocities (<170 cm/s, normal; ≥170 but <200 cm/s, conditional; and ≥200 cm/s, abnormal).⁹ Children with abnormal or conditional TCD in any vessel were considered high risk.⁸ For children who underwent transfusion therapy, TCD readings just before the transfusion were analyzed. The TCDs were analyzed by an investigator (S.T.M.) masked to PSG and PSQ results.

A thorough medical record review was conducted for each patient for a history of hydroxyurea or transfusion therapy, acute chest syndrome, tonsillectomy and/or adenoidectomy, treatment with continuous positive airway pressure, prematurity, or asthma and current height and weight. Body mass index was calculated as weight in kilograms divided by height in meters squared and compared with standard percentiles for age and sex. Nonclinical factors analyzed included age, sex, race, and socioeconomic status, as determined by insurance and maternal educational level.

The prevalence of snoring, SDB as determined by PSQ score greater than 0.33, and SDB as determined by positive PSG findings, along with the corresponding 95% CIs, were determined for the entire sample and for the sample excluding children who were previously treated for SDB. Comparison of TCD velocity between nonsnorers, snorers with negative PSG findings, and snorers with positive PSG findings and between nonsnorers, snorers with negative PSQ scores, and snorers with positive PSQ scores was performed using the Kruskal-Wallis test. Comparison of high-risk TCD velocities between the above-mentioned groups was performed using the generalized Fisher exact test. Both of these analyses were also performed for the entire sample and for untreated children. The following potential predictors of SDB were compared between children with positive PSG findings and the rest of the sample and between children with positive PSQ scores and children with negative PSQ scores using the Fisher exact test: sex, prematurity, asthma, percentile body mass index, transfusion protocol, hydroxyurea therapy, insurance, and maternal educational level. The Mann-Whitney test was used to compare age between the aforementioned groups. We used SAS, version 9.2 (SAS Institute Inc) for statistical analysis. P < .05 was considered statistically significant.

Sixty-nine patients were offered participation into the study. Five patients declined, so a total of 64 patients were included. Patient demographic characteristics and medical history are presented in Table 1. One patient (2%) had a prior adenoidectomy for treatment of otitis media. No patient had been prescribed nasal continuous positive airway pressure for treatment of SDB.

The prevalence of snoring was 37.5%, the prevalence of SDB as documented by positive PSG findings was 23.7%, and the prevalence of SDB as defined as a positive PSQ score was 21.9% (Table 2). When the analysis was performed excluding the 9 patients who were previously treated for SDB, the prevalence estimates were similar. For the 14 studies with positive PSG findings, the median apnea-hypopnea index was 4.7 (range, 1.5-71.2) and the median lowest oxygen saturation was 83% (56%-97%).

Fifty-seven patients had TCD, 56 studies were interpretable, and 1 study was not adequate for interpretation. There was no significant difference in TCD velocity or number of patients with high-risk TCD between nonsnorers and snorers, regardless of their PSG findings or PSQ scores or whether previously treated patients were included (Table 3).

On bivariate analysis of the entire group, positive PSG or PSQ findings were not associated with traditional risk factors for pediatric SDB, except there was a trend toward association of abnormal PSG with asthma and of PSQ scores with obesity. Therapy with regular transfusion or hydroxyurea was not a predictor of SDB (Table 4). Age was also not a significant predictor because the mean (SD) age of children with positive PSG findings was 8.0 (3.2) years compared with 8.3 (4.0) years for the remainder of the children (P = .85), and the age of children with positive PSQ scores was 8.5 (3.6) years compared with 8.3 (4.0) years for children with negative PSQ scores (P = .81). Because of the small sample size and lack of significant predictors on bivariate analysis, multivariate analysis was not performed.

All children with positive PSG findings (n = 14) were referred for consideration of AT. To date, 6 children have undergone AT, 2 have undergone adenoidectomy, the parents of 2 children refused the treatment, and 4 children with mild SDB (apnea-hypopnea index <3.7) are being followed up. Three patients had follow-up TCD (at 6, 4, and 6 months after AT), and TCD velocity was unchanged (138 cm/s initial vs 138 cm/s postoperative, 183 cm/s initial vs 181 cm/s postoperative, and 163 cm/s initial vs 165 cm/s postoperative, respectively). Two additional patients received regular transfusions because of newly discovered abnormal TCD velocity.

In this cross-sectional study of children with SCD (predominantly type SS), the prevalence of snoring was 37.5% and of SDB as documented by PSG was 23.7%. An additional 9 children (14%) had prior treatment for SDB. Using similar methods, Samuels et al⁶ found a prevalence of SDB of 36%, and Salles et al⁵ found a prevalence of snoring of 44.7% and of SDB of 10.6% in consecutive patients with SCD attending routine clinic visits. Possible mechanisms for the increased risk and severity of SDB in SCD are compensatory adenotonsillar hyperplasia after splenic infarction or anatomic changes in the upper airway secondary to bone marrow hyperplasia.^3,14

Sleep-disordered breathing was determined by a positive score on the PSQ and by the criterion standard of overnight PSG. Only snoring children were referred for PSG because SDB is rare in nonsnorers. The PSQ has been shown to have a sensitivity of 81% and a specificity of 87% compared with PSG and is useful for research purposes, particularly when PSG is not feasible.¹³ For SDB estimates, the use of selected PSG in snoring children and the use of PSQ scores yielded similar prevalence estimates, although a sensitivity analysis of PSQ in predicting results of PSG could not be performed because not all children underwent PSG.

Although cerebral infarction is the most devastating complication of SCD, its mechanisms are poorly understood. Narrowing and occlusion of the intracranial carotid and middle cerebral arteries, anemia, hypoxemia, impaired cerebral autoregulation, increased blood viscosity, lung disease, and acute medical events resulting in reduced baseline hemoglobin all play a role.^10,15 It has been proposed that recurrent hypoxemia from OSA contributes to vessel abnormalities and vaso-occlusion in small cerebral vessels just as it occurs in other organs during episodes of vaso-occlusive pain crises.¹⁶ Diurnal hypoxemia has been shown to be associated with TCD velocity in patients with SCD.¹¹

Notwithstanding theoretical considerations, there have been few published studies supporting the role of SDB in the pathogenesis of cerebrovascular disease. Several case reports^16,17 describe children with both strokes and OSA. Bader-Meunier et al¹⁸ described 2 children with SCD and elevated TCD whose TCDs normalized after treatment of SDB (1 with AT and 1 with adenoidectomy). Kirkham et al¹⁹ screened 95 children with SCD with overnight pulse oximetry and observed them for the development of central nervous system events for a median period of 6 years. Low mean oxygen saturation and a high proportion of the night with oxygen saturation less than 90% were predictive of future central nervous system events, but dips in oxygen saturation suggestive of OSA and a history of AT were not. Polysomnography was not performed, and subsequent studies have shown that nocturnal desaturation does not correlate with apnea index or respiratory disturbance index in children with SCD.²⁰ In a matched cohort study²¹ of 256 children with SCD who had undergone AT compared with 512 children who had not, the treated children had significantly reduced rates of visits and service costs over time for OSA, stroke, and transient ischemic attacks but not for vaso-occlusive pain, acute chest syndrome, or pneumonia for 3 years after surgery.

We found no association between elevated TCD velocity or high-risk TCD and SDB. The study was cross-sectional, 9 children (14%) had prior treatment for SDB, and 6 (9%) were already receiving a regular transfusion protocol. We analyzed the data both including and excluding patients previously treated for SDB to be sure that including children previously treated was not falsely lowering the TCD velocities. We also used the most abnormal TCD preceding transfusion therapy for children receiving the transfusion protocol because transfusions are known to lower TCD velocity. Similarly, we excluded children with a history of stroke because the number of patients who had a stroke was small and most had not had TCD screening before stroke onset and initiation of transfusion therapy. Ideally, our study should be performed prospectively by enrolling children with new diagnoses of SCD and performing serial TCD and PSG, but this would require extensive time and resources. Three of our patients had postoperative TCD, and the velocities were unchanged.

Analysis for association of the usual predictors of SDB in otherwise healthy children (younger age, male sex, prematurity, asthma, overweight/obesity, and low socioeconomic status) did not demonstrate any significant associations in our patients with SCD. Risk factors may differ in children with SCD in whom adenotonsillar hyperplasia is common, and very few children are overweight while many are underweight. As reported in otherwise healthy children, improved growth velocity has been found after AT in children with SCD.³ Daniel et al²² found that living below the poverty line was correlated with SDB score on the Children's Sleep Habits Questionnaire in 54 children with SCD.

Hill et al²³ found significantly elevated TCD velocities in 31 healthy children with mild SDB (apnea-hypopnea index <5) compared with 17 age-similar control patients; normal TCD velocity ranges from 75 to 95 cm/s in healthy children. The children with SDB had a mean TCD velocity of 120 cm/s compared with healthy controls who had a mean TCD velocity of 84 cm/s. In a follow-up study²⁴ by the same group, TCD velocity significantly decreased after AT in 19 children, whereas it increased 12 months later in 14 age-similar controls who did not undergo surgery. Although for the purposes of determining stroke risk, less than 170 cm/s was designated as “normal” for the Stroke Prevention Trial in Sickle Cell Anemia, the mean flow velocity even in this low-risk group was 129 cm/s in published pilot data, which are considerably higher than normal velocities in children unaffected by SCD.⁸ It may be that the higher baseline velocities negated the effect of AT on TCD findings and/or that cerebral blood flow velocities are nearly maximized at baseline in children with SCD²⁵; we saw no difference from before to after operation in our 3 patients who had postoperative screening performed. Measures of neurocognition also significantly improved after surgery, although a direct causative relationship has yet to be demonstrated; we did not include neuropsychologic testing in our study.

The strengths of our study are the use of 2 hospital-based pediatric hematology clinics for patient recruitment, validated assessments for SDB and cerebrovascular disease, and masking. Weaknesses include the cross-sectional nature of the study, the use of a convenience sample, loss to follow-up/refusal (5 patients [20%] did not have PSG and 7 [11%] did not have TCD), and a small sample size. Although it was a convenience sample, more than 90% of eligible patients were approached for potential recruitment at the State University of New York Downstate Medical Center and 75% at Kings County Hospital Center. It is possible that a larger sample size would have resulted in more significant associations and would have allowed a multivariate analysis of potential risk factors for SDB.

In conclusion, in this cross-sectional study, the prevalence of snoring in children with SDB was 37.5% and the prevalence of SDB was 23.7%. There was no association of SDB with elevated TCD velocity or with high-risk TCD. A larger prospective study is needed to confirm these findings.

Correspondence: Nira A. Goldstein, MD, MPH, Division of Pediatric Otolaryngology, State University of New York Downstate Medical Center, 450 Clarkson Ave, Campus Box 126, Brooklyn, NY 11203 (ngoldstein@downstate.edu).

Submitted for Publication: May 9, 2011; final revision received August 10, 2011; accepted September 4, 2011.

Author Contributions: Drs Goldstein, Keller, Rao, and Miller and Ms Rey had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Goldstein, Rao, Weedon, and Miller. Acquisition of data: Keller, Rey, Dastgir, and Mironov. Analysis and interpretation of data: Goldstein, Keller, Rao, Weedon, and Miller. Drafting of the manuscript: Goldstein and Keller. Critical revision of the manuscript for important intellectual content: Goldstein, Keller, Rey, Rao, Weedon, Dastgir, Mironov, and Miller. Statistical analysis: Weedon. Administrative, technical, and material support: Keller and Rey. Study supervision: Goldstein, Rao, and Miller.

Previous Presentation: This study was presented at the American Society of Pediatric Otolaryngology meeting; April 30, 2011; Chicago, Illinois.

Additional Contributions: We thank Katharina D. Graw-Panzer, MD, for reviewing the polysomnograms and Michael Chan, MD, and Bernard Koliskor, MD, for their help with patient recruitment.