Upper Airway Lymphoid Tissue Size in Children With Sickle Cell Disease

Abstract

Background:

The prevalence of obstructive sleep apnea syndrome (OSAS) is higher in children with sickle cell disease (SCD) as compared with the general pediatric population. It has been speculated that overgrowth of the adenoid and tonsils is an important contributor.

Methods:

The current study used MRI to evaluate such an association. We studied 36 subjects with SCD (aged 6.9 ± 4.3 years) and 36 control subjects (aged 6.6 ± 3.4 years).

Results:

Compared with control subjects, children with SCD had a significantly smaller upper airway (2.8 ± 1.2 cm³ vs 3.7 ± 1.6 cm³, P < .01), and significantly larger adenoid (8.4 ± 4.1 cm³ vs 6.0 ± 2.2 cm³, P < .01), tonsils (7.0 ± 4.3 cm³ vs 5.1 ± 1.9 cm³, P < .01), retropharyngeal nodes (3.0 ± 1.9 cm³ vs 2.2 ± 0.9 cm³, P < .05), and deep cervical nodes (15.7 ± 5.7 cm³ vs 12.7 ± 4.0 cm³, P < .05). Polysomnography showed that 19.4% (seven of 36) of children with SCD had OSAS compared with 0% (zero of 20) of control subjects (P < .05) and that in children with SCD the apnea-hypopnea index correlated positively with upper airway lymphoid tissues size (r = 0.57, P < 001). In addition, children with SCD had lower arterial oxygen saturation nadir (84.3% ± 12.3% vs 91.2% ± 4.2%, P < .05), increased peak end-tidal CO₂ (53.4 ± 8.5 mm Hg vs 42.3 ± 5.3 mm Hg, P < .001), and increased arousals (13.7 ± 4.7 events/h vs 10.8 ± 3.8 events/h, P < .05).

Conclusions:

Children with SCD have reduced upper airway size due to overgrowth of the surrounding lymphoid tissues, which may explain their predisposition to OSAS.

Sleep-disordered breathing (SDB), a condition associated with abnormalities in respiratory gas exchange during sleep (particularly intermittent oxyhemoglobin desaturation), has been implicated as a possible risk factor for deleterious cerebrovascular complications in children with sickle cell disease (SCD).^1‐3 Although the prevalence of SDB in these children has not been well established, estimates range from 5% to 79%^4‐10 and are far greater than the prevalence of 1% to 4% in children without SCD.¹¹ In addition, the pathophysiology of SDB in these children is not well understood. Possible mechanisms include hypoventilation due to chronic lung disease,^12,13 obstructive sleep apnea syndrome (OSAS)⁶ (a disorder characterized by recurrent events of partial or complete upper airway obstruction during sleep¹⁴ that is associated with distinct neurocognitive deficits and cardiovascular morbidities¹⁵), left ventricular diastolic dysfunction,¹⁶ pulmonary hypertension,¹⁷ and the presence of dyshemoglobins.¹⁸

This study was part of a large investigation at the Children’s Hospital of Philadelphia (CHOP) on the prevalence, contributory mechanisms, and pulmonary and vascular consequences of oxyhemoglobin desaturations in children with SCD-hemoglobin. In this particular study, we aimed to determine the possible anatomic changes in the upper airway that may predispose to SDB and particularly OSAS.

In children without SCD, adenotonsillar hypertrophy is the most common cause of OSAS.^19‐21 Children with SCD may be predisposed to the development of adenoidal and tonsillar hypertrophy due to a compensatory response for their commonly described functional asplenia.^4,6,9 Thus, we hypothesized that children with SCD may have excess upper airway lymphoid tissue proliferation leading to a decrease in their upper airway size. Such alterations may in turn result in abnormal polysomnographic findings and increase the prevalence of SDB and OSAS in this population. If such an association exists, greater attention to diagnosis and management of compromised airway size in this population would be warranted to reduce oxyhemoglobin desaturation during sleep. To this end, we used MRI to delineate the size of the upper airway and surrounding lymphoid tissues and overnight polysomnography to evaluate for SDB and/or OSAS in unselected children with SCD as compared with control subjects.

Materials and Methods

Subjects

The study was approved by the institutional review boards of CHOP (IRB#07-005188) and the Children’s Hospital at Montefiore (CCI#2007-941). Informed assent was obtained from each child > 7 years old, and informed consent was obtained from a parent/guardian of each child.

Subjects With SCD:

Children were recruited from the Comprehensive Sickle Cell Center at CHOP. Inclusion criteria were as follows: (1) SCD-hemoglobin; (2) age 2 to 21 years; (3) intact adenoid and tonsils; (4) steady state, defined as a period of at least 3 months since the last RBC transfusion and at least 4 weeks since the last acute chest syndrome or painful episode. Exclusion criteria were as follows: (1) treatment with hydroxyurea within the past 3 months; (2) chronic lung disease unrelated to SCD, other than asthma; (3) chronic transfusion protocol.

Control Subjects:

Control subjects were selected from patients who underwent a head or neck MRI at CHOP for medical indications such as concussion, headache, or seizures. Control subjects were matched to subjects with SCD by age, sex, ethnicity, weight, and height. Inclusion criteria were (1) normal growth and development, (2) intact adenoid and tonsils. Exclusion criteria included (1) evidence of a brain tumor or a seizure disorder requiring therapy, (2) genetic disorders associated with any craniofacial anomaly, (3) chronic respiratory disease other than asthma, (4) history of OSAS.

MRI

Studies were performed in the Department of Radiology at CHOP. For children < 7 years of age, studies were performed under sedation with IV pentobarbital 2 to 6 mg/kg until sleep was achieved; a maximum of 200 mg was administered. All subjects were monitored continuously by pulse oximetry and observed by an anesthesiologist throughout the study until recovery (~1 h).

MRI was performed with a 1.5T Siemens Vision system. Images were acquired using a commercially available head coil. Axial and sagittal sequential T1-weighted (TR650/TE14) and T2-weighted (TR6000/TE90) images with 3-mm slice thickness and 1 NEX were obtained from the orbital cavity to the larynx and from the midline bilaterally, respectively, as previously described.²¹

Image Processing and Upper Airway Measurements

The acquired MRI studies were anonymized and converted to a multidimensional version of the digital imaging and communications in medicine (DICOM) format. These were transferred via the CHOP’s picture archiving system to a workstation at the Division of Respiratory and Sleep Medicine at the Children’s Hospital at Montefiore. Image analysis was performed by a blinded scorer using AMIRA, version 4.1.1, software (Visage Imaging, GmbH), and using intensity threshold after normalization.

Volumetric Measurements

The volumes of the following structures were determined as follows:

1. Airway: The upper airway including the nasopharynx, defined as the region located superior to the level of the soft palate and continuous anteriorly, through the choanae, with the nasal cavities; oropharynx, defined as the region located between the level of the soft palate and the larynx, communicating anteriorly with the oral cavity, and having the posterior one-third of the tongue as its anterior border; and the hypopharynx, defined as the region posterolateral to the larynx, and communicating with the cavity of the larynx through the auditus and included the pyriform recesses and the valleculae.

2. Lymphoid tissues: Adenoid, combined palatine tonsils, combined retropharyngeal nodes (defined as lymph nodes located between the internal carotid arteries from the base of the skull to the hyoid bone), and the combined deep cervical lymph nodes (defined as level 2 nodes, located along the internal jugular vein from base of the skull to the level of the hyoid bone).

Polysomnography

Overnight polysomnography was performed in the sleep laboratory at CHOP. The following parameters were recorded (using Embla): EEGs (C4/A1, C3/A2, O1/A2, O2/A1), right and left electrooculogram, submental and tibial electromyograms, chest and abdominal wall movement (Respitrace Systems; Ambulatory Monitoring Inc), ECG, end-tidal CO₂ (etco₂) by capnography (Novametrix 7000; Novametrix), airflow by nasal pressure (Pro-Tech) and three-pronged thermistor (Nihon Kohden), arterial oxygen saturation (Spo₂) and pulse waveform (Masimo), and infrared video. Sleep staging and respiratory events were consistent with the American Academy of Sleep Medicine pediatric scoring rules.²² Accordingly, central apneas, obstructive apneas, and obstructive hypopneas were scored. An obstructive apnea-hypopnea index (AHI) was calculated as the number of obstructive apneas and hypopneas per hour. The reported Spo₂ nadir was the lowest oxygen saturation measured during polysomnography and not limited to an apnea or hypopnea event. We considered the diagnosis of OSAS when the AHI was ≥ 1.5/h.^23‐25

Sleep Questionnaire

Since not all control subjects underwent polysomnography, we used a validated questionnaire developed by Brouilette et al²⁶ to assess the likelihood of OSAS in these subjects. Accordingly, no subject with score < −1 would be expected to have OSAS; a score between −1 and 3.5 is considered indeterminate, and a score > 3.5 is considered highly predictive of OSAS.

Data Analysis

Statistical analysis was conducted using SPSS, version 18 (SPSS Inc). Means and SDs were used to summarize continuous variables. For comparisons between the groups for MRI data demographics, anthropometrics, and polysomnography data, we used a two-tailed unpaired t test and χ² test as appropriate. Pearson correlations were derived between AHI and upper airway lymphoid tissues within the SCD group. A P value < .05 was considered significant.

Results

We studied 36 children with SCD with a mean age of 6.9 ± 4.3 years (2.0-16.8 years) and 36 control subjects with a mean age of 6.6 ± 3.4 years (2.2-15.8 years). Subjects with SCD were similar to control subjects in age, ethnicity, sex, height, and weight (Table 1). However, their mean BMI z score was significantly lower (Table 1).

Table 1.

—Demographics and Anthropometric Measures

Measure	SCD (n = 36)	Control Subjects (n = 36)	P Value
Age, y	6.9 ± 4.3	6.6 ± 3.4	NS
Range, y	2.0-16.8	2.2-15.8
Ethnicity: black, No.	36	36	NS
Sex, male, %	55.6	55.6	NS
Height, cm	117.6 ± 24.8	117.4 ± 21.8	NS
Weight, kg	23.5 ± 13.3	25.2 ± 11.7	NS
BMI z score	−0.4 ± 1.2	0.6 ± 1.0	< .05

Data are displayed as mean ± SD unless otherwise noted. NS = not significant; SCD = sickle cell disease.

Upper Airway Volumetric Measurements

Volumetric analysis based on 3-mm axial images of the upper airway is shown in Table 2. Upper airway lymphoid tissue analysis of an 11.8-year-old male subject with SCD with OSAS, using AMIRA, is presented graphically in Figure 1. Figure 2 depicts a three-dimensional rendering of the face, upper airway, and lymphoid tissues of the same subject.

Table 2.

—Airway and Lymphoid Tissues Volumes

Area Measured	SCD (n = 36)	Control Subjects (n = 36)	% Difference	P Value
Airway	2.8 ± 1.2	3.7 ± 1.6	−24.3	< .01
Lymphoid tissues
Adenoid	8.4 ± 4.1	6.0 ± 2.2	40.0	< .01
Tonsils	7.0 ± 4.3	5.1 ± 1.9	37.3	< .01
Retropharyngeal  nodes	3.0 ± 1.9	2.2 ± 0.9	36.4	< .05
Deep cervical  nodes	15.7 ± 5.7	12.7 ± 4.0	23.6	< .05

Data are displayed a mean ± SD. Units are cm³. % Difference = percent mean volume difference. See Table 1 legend for expansion of abbreviation.

Figure 1.

Upper left, Surface rendering of the head and neck with three-dimensional reconstruction of the adenoid (magenta), tonsils (orange), retropharyngeal nodes (red), and deep cervical lymph nodes (green), of an 11.8-year-old male subject with sickle cell disease with obstructive sleep apnea syndrome using AMIRA software. Upper right, Axial T2-weighted image at the nasopharyngeal level outlining the adenoid (magenta). Lower left, Coronal reconstructed image outlining the adenoid (magenta) and tonsils (orange). Lower right, Midsagittal reconstructed image outlining the adenoid (magenta) and tonsils (orange).

Figure 2.

Surface rendering of the head and neck and three-dimensional reconstruction of the upper airway and lymphoid tissues of the subject shown in Figure 1: airway (light blue), adenoid (magenta), tonsils (orange), retropharyngeal nodes (red), and deep cervical lymph nodes (green).

Airway:

We noted a significantly smaller upper airway in the SCD group. In comparison with control subjects, subjects with SCD had an upper airway volume of 2.8 ± 1.2 cm³ compared to 3.7 ± 1.6 cm³ in control subjects (P < .01).

Lymphoid Tissues:

All lymphoid tissues surrounding the upper airway, including adenoid, tonsils, and retropharyngeal nodes, as well as the deep cervical nodes, were significantly larger in children with SCD as compared with control subjects (Table 2). For the SCD groups, we noted the following correlations between AHI and the various lymphoid tissues: adenoid, r = 0.71 (P < .001); retropharyngeal nodes, r = 0.61 (P < .001); tonsils, r = 0.1 (P = not significant [NS]), and deep cervical nodes, r = 0.1 (P = NS). A positive correlation was noted between AHI and the combined volume of the lymphoid tissues surrounding the upper airway (tonsils, adenoid, and retropharyngeal nodes; r = 0.57, P < .001) (Fig 3).

Figure 3.

Correlation between apnea-hypopnea index and the combined upper airway lymphoid tissue volume (cm³) in the group with sickle cell disease.

Polysomnography

Polysomnography data were available for all 36 subjects with SCD and for 20 control subjects and are shown in Table 3. Abnormalities in gas exchange in SCD, including lower baseline Spo₂, lower Spo₂ nadir, increased baseline etco₂, and increased peak etco₂, are suggestive of SDB in this group. In addition, subjects with SCD exhibited differences in sleep quality, including decreased sleep efficiency and increased arousals, as compared with control subjects.

Table 3.

—Polysomnography

Measure	SCD (n = 36)	Control Subjects (n = 20)	P Value
Total sleep time, h	7.3 ± 1.2	7.7 ± 0.8	NS
Sleep efficiency, %	83.7 ± 12.4	90.4 ± 5.3	< .05
Arousal index, events/h	13.7 ± 4.7	10.8 ± 3.8	< .05
Baseline Spo2, %	95.3 ± 2.9	97.1 ± 0.9	< .05
Spo2 nadir, %	84.3 ± 12.3	91.1 ± 4.2	< .05
Baseline etco2, mm Hg	43.0 ± 3.1	37.5 ± 4.6	< .001
Peak etco2, mm Hg	53.4 ± 8.5	42.3 ± 5.3	< .001
Obstructive apnea index, events/h	0.7 ± 2.0	0.2 ± 0.3	NS
AHI	1.9 ± 4.7	0.4 ± 0.3	NS
OSAS (AHI ≥ 1.5)	7 of 36	0 of 20	< .05

Data are displayed as mean ± SD. AHI = apnea-hypopnea index; etco₂ = end-tidal CO₂; OSAS = obstructive sleep apnea syndrome; Spo₂ = arterial oxygen saturation. See Table 1 legend for expansion of other abbreviations.

In regard to OSAS, on average the SCD group did not have more obstructive apneas or obstructive hypopneas. However, seven of 36 (19.4%) of the SCD group had an AHI elevated above a suggestive threshold for OSAS as compared with zero of 20 (0%) in the control group (P < .05). These seven subjects with OSAS had a mean obstructive apnea index of 3.0 ± 3.9 events/h (median, 2.5 events/h), a mean AHI of 7.8 ± 8.9 events/h (median, 5.9 events/h), a mean Spo₂ nadir of 80.4% ± 8.1% (median, 83%), a mean peak etco₂ of 52.7 ± 4.3 mm Hg (median, 54.6 mm Hg), and a mean arousal-awakening index of 14.9 ± 5.1 events/h (median, 15.2 events/h).

Sleep Questionnaire

All control subjects had an OSAS questionnaire score < −1, suggesting that none had evidence of the disorder. In addition, scores in control subjects who had polysomnography were similar to those who did not (−3.0 ± 0.8 vs −3.1 ± 0.9, respectively; P = NS).

Discussion

The current study is the first, to our knowledge, to quantify the upper airway and surrounding lymphoid tissues in children with SCD. Our findings suggest that SCD is associated with a generalized increase in size of upper airway lymphoid tissues and concomitant decrease in upper airway size. We believe the current analysis can explain the high occurrence of OSAS in the subjects with SCD in our study and complements previous studies demonstrating a much higher prevalence of OSAS in subjects with SCD compared with the general population.^4‐10

A few methodologic issues deserve comment. First, MRI is considered a reliable and accurate tool to evaluate the upper airway and surrounding soft tissues. However, to minimize movement artifact during imaging, light sedation was provided to all children < 7 years of age. It is possible that sedation altered upper airway dimensions in these children. However, this effect was indeed controlled by studying a similar number of sedated children in each group. Sedation should not have affected the volumetric measurements of the lymphoid tissues. Second, in spite of our attempt to match control subjects to subjects with SCD by demographics and anthropometrics, BMI z score was significantly lower in subjects with SCD. This is probably due to suboptimal growth described in children with SCD.^27,28 We believe, however, that this finding of larger lymphoid tissues in the SCD group despite having a lower BMI z score further enhances our findings. Third, we studied subjects at steady state and excluded those requiring chronic blood transfusion or hydroxyurea or those with episodes of acute chest syndrome or pain episodes within 4 weeks of the study.

Our hypothesis considered existence of general lymphoid hypertrophy in proximity to the upper airway in children with SCD. Therefore, our analysis was not limited to the adenoid and tonsils but also included the retropharyngeal nodes. The retropharyngeal nodes are located posterior to the pharyngeal constrictor muscles and between the internal carotid arteries from the base of the skull to the hyoid bone. These nodes may contribute to upper airway restriction and risk for OSAS when they are enlarged, in a similar way to the adenoid and tonsils.²⁹ This point is not well documented in the literature since these nodes are not visible on examination and are not routinely evaluated by radiographic measures. Our findings of larger adenoid, tonsils, and retropharyngeal nodes restricting the upper airway could explain the propensity of children with SCD to have OSAS. This is also supported by the demonstration of a positive correlation between the AHI and the size of lymphoid tissues surrounding the upper airway.

The large deep cervical nodes noted in these subjects should not directly impact upper airway size because of their more distant location from the airway. However, the finding that the size of deep cervical nodes is also increased in SCD lends support to the observation of overall lymphoid hypertrophy in these children.

The main polysomnographic findings in our study included abnormalities in gas exchange in subjects with SCD; these findings are not considered specific and may represent abnormalities in ventilation/perfusion ratio, hypoventilation, or abnormalities in gas diffusion. However, the more specific finding is of OSAS in 19% (seven of 36), which supports other studies evaluating the increased incidence of SDB and OSAS in this population.^4‐10

Three important weaknesses related to our study should be mentioned. First, control subjects were excluded if any had a history of OSAS and/or previously underwent adenotonsillectomy. Thus, a possible selection bias may have been introduced by such criteria. However, we do not think this should significantly affect our results, assuming a background prevalence of 2% of children in this age group with OSAS.

Second, not all control subjects who had an upper airway MRI agreed to have polysomnography. For this reason, we introduced a standardized sleep questionnaire to exclude the possibility of OSAS in any of the 16 additional control subjects. Interestingly, a subgroup analysis comparing the 36 subjects with SCD to the 20 control subjects who agreed to have polysomnography demonstrated similar findings to our primary analysis. The SCD group had a smaller upper airway (P = .0005) and larger tonsils (P = .03). However, despite the fact that the adenoid and deep cervical nodes were identical in size to the primary analysis, the P values only approached significance (P = .08 and P = .06, respectively). The latter finding may reflect the smaller sample size within this subanalysis.

Third, in this study we focused on upper airway lymphoid tissues as the main cause for OSAS in SCD. However, it is possible that children with SCD have other anatomic abnormalities, such as those affecting their craniofacial structure particularly due to extramedullary hematopoiesis,³⁰ and/or functional factors that may increase upper airway collapsibility,³¹ thereby increasing their risk for SDB or OSAS. Thus, additional studies are warranted to elucidate such potential contributors in this population.

Although the reason for upper airway lymphoid hypertrophy in SCD remains unknown, it has been speculated that it may result from hematopoietic lymphoid compensation secondary to functional asplenia.^4,6,9 In this regard, we would like to point out that in SCD, chronic hemolysis and vasoocclusive events are associated with local and systemic inflammatory responses to oxidative stress, ischemia-reperfusion injury, and release of NO.^32‐34 Similar pathways of oxidative stress reaction have been noted in children without SCD who have SDB and/or OSAS with adenotonsillar hypertrophy.^35‐39 Thus, we propose that similar mechanisms of inflammatory response that occur in children with SCD may at least partly explain their upper airway lymphoid hypertrophy.

In conclusion, we suggest that children with SCD should be screened for OSAS by an appropriate questionnaire and physical examination because of their risk of developing adenotonsillar hypertrophy. Adenotonsillectomy should be considered once diagnosis of OSAS is confirmed to reduce the deleterious effects of the disorder in this particular group.

Abbreviations

AHI

apnea-hypopnea index

CHOP

Children’s Hospital of Philadelphia

etco₂

end-tidal CO₂

NS

not significant

OSAS

obstructive sleep apnea syndrome

SCD

sickle cell disease

SDB

sleep-disordered breathing

Spo₂

arterial oxygen saturation