PDF
Citation
Print
INTRODUCTION
Obstructive sleep apnea (OSA) is a common disorder characterized by repeated episodes of complete or partial upper airway obstruction during sleep [1,2]. Patients with OSA are frequently comorbid with cardiovascular, cerebrovascular, and metabolic diseases and have increased mortality rates [1,2]. Acromegaly is a rare disorder that is typically caused by a growth hormone (GH)-secreting pituitary adenoma [3]. Patients with acromegaly have increased mortality rates, predominantly owing to cardiometabolic comorbidities [3]. OSA is highly prevalent in patients with acromegaly, and 20% to 100% of these patients are affected by OSA [4]. Excess circulating GH and insulin-like growth factor 1 (IGF-1) induce anatomical changes in the craniofacial bones and pharyngeal soft tissue swelling, resulting in an increased risk of upper airway obstruction during sleep [5,6].
Although OSA is common in patients with acromegaly, there is currently no reliable and valid screening tool to screen for the presence of OSA in patients with acromegaly. In addition, although several previous studies have reported that the effective treatment of acromegaly is associated with the improvement of OSA [4], these studies have been limited because of the small sample size and heterogeneous population undergoing medical and/or surgical treatment of acromegaly. Here, we aimed to examine the predictive performance of the Snoring, Tiredness, Observed apnea, high blood Pressure-Body mass index, age, Neck circumference, and Gender (STOP-Bang) questionnaire [7] as a screening tool for OSA. We also aimed to evaluate the effects of endoscopic transsphenoidal surgery on the improvement of OSA and its associated factors in patients with acromegaly.
METHODS
Study design and participants
In this observational before–after study, we prospectively enrolled adult patients with acromegaly who were scheduled for endoscopic transsphenoidal surgery between January 2018 and May 2022 at Seoul National University Hospital (Trial Registration: NCT03526016). We excluded patients with a history of pituitary surgery for acromegaly, those who were unable to undergo transsphenoidal surgery, those currently being treated for sleep apnea, and those who were unable to provide informed consent. All measurements were conducted when the participants were admitted for a baseline work-up for acromegaly before surgery and postoperative surveillance approximately 3 to 6 months after surgery.
This study was approved by the Institutional Review Board of Seoul National University Hospital (H-1802-032-919). All participants provided written informed consent, and this study was conducted in accordance with the tenets of the Declaration of Helsinki.
Hormonal and clinical assessments for acromegaly
The diagnosis of acromegaly was based on a serum GH concentration >1 ng/mL after a 75-g oral glucose tolerance test (OGTT) and an elevated serum IGF-1 level above the age- and sex-specific reference range [6,8]. Biochemical remission was defined as a suppressed serum GH level of <1 ng/mL during the OGTT and a serum IGF-1 level within the age- and sex-specific reference range at 3 to 6 months after surgery [3,8]. Details of the biochemical assessment of other hormones have been described previously [8]. Briefly, adrenocorticotropic hormone (ACTH) deficiency was defined as a peak cortisol level of <18 μg/dL after a short Synacthen test or a low or normal ACTH level with a low morning cortisol level (≤5 μg/dL). Thyroid-stimulating hormone (TSH) deficiency was defined as low or normal TSH levels (reference range, 0.4 to 4.1 μIU/mL) with low free thyroxine level (<0.7 ng/dL). Gonadotropin deficiency was defined as low or normal follicle-stimulating hormone and luteinizing hormone levels with a low testosterone or estradiol level. Maximal tumor size was defined as the largest diameter of the three spatial dimensions measured using T1-weighted contrast-enhanced magnetic resonance imaging (MRI) of the pituitary. Detailed descriptions of the endoscopic transsphenoidal surgery procedures have been presented in previous studies [9,10]. Weight gain was defined as an increase of ≥3% compared with before surgery, weight loss as a decrease of ≥3%, and stable weight as change in weight between these values.
Respiratory polygraphy and assessment for OSA
A type 3 sleep study using respiratory polygraphy (Embla-Embletta, Natus, Pleasanton, CA, USA) was conducted from 10:00 PM to 7:00 AM of the next day during hospitalization. Respiratory polygraphy measurements included airflow, oxygen saturation (SpO2), and respiratory effort using an oronasal thermal sensor and nasal pressure transducer, a pulse oximeter, and dual thoracoabdominal respiratory inductance plethysmography belts, respectively. Each study was automatically reviewed, followed by a manual review according to the American Academy of Sleep Medicine guidelines [11]. Apnea was defined as ≥90% decrease in the peak signal excursion from the pre-event baseline using an oronasal thermal sensor for ≥10 seconds. Nasal pressure sensors were used as a backup if the signal of the oronasal thermal sensor was unreliable. Hypopnea was defined as ≥30% decrease in peak signal excursion from the pre-event baseline using a nasal pressure sensor for ≥10 seconds and as ≥ 3% oxygen desaturation from the pre-event baseline. The monitoring time was calculated as the total recording time minus the periods of artefacts and the time the patient was awake, as determined by a body position sensor, respiratory pattern, or patient note. The respiratory event index (REI) was calculated as the number of apneas plus hypopneas divided by the monitoring time [12]. The REI is a surrogate for the apnea-hypopnea index (AHI), which is analogous to the AHI used in type 1 polysomnography recording sleep. A patient with REI ≥5/hour was diagnosed as having OSA. REI ≥15/hour was classified as moderate-to-severe, and REI ≥30/hour was classified as severe [13]. The oxygen desaturation index (ODI) was defined as episodes of ≥3% oxygen desaturation per hour during monitoring. The sleep time spent with SpO2 <90% (TS90) and the mean and lowest SpO2 values were also recorded. Patients with invalid respiratory polygraphy results were excluded from analysis.
The STOP-Bang questionnaire was used to evaluate the risk of OSA [7]. The STOP-Bang questionnaire was initially developed to screen patients with undiagnosed OSA preoperatively and was further validated in numerous studies [14]. The questionnaire comprises eight yes/no questions (Snoring, Tiredness, Observed apnea, high blood Pressure, body mass index [BMI] >35 kg/m2, age >50 years, neck circumference >40 cm, and male gender) [7]. Using a cutoff score of 3, it has an accuracy of 84% for detecting OSA of any severity and 93% for moderateto-severe OSA [7]. The questionnaire’s Korean translation was validated [15]. Based on the definition of obesity according to the World Health Organization and Asian-Pacific guidelines [16], a BMI cutoff value of 30 kg/m2 rather than 35 kg/m2 was used in this study. The Epworth Sleepiness Scale (ESS) was used to assess the propensity for excessive daytime sleepiness [17]. It consists of eight questions describing situations, each of which receives a score of 0 to 3. A score of >10 is consistent with excessive sleepiness. The clinical significance threshold of this questionnaire is 2 points [18]. The Korean version of the questionnaire was also validated [19].
Statistical analyses
Categorical variables are presented as counts and percentages. Continuous variables are presented as medians with interquartile ranges. The missing values of neck circumference, Friedman tongue position, and tonsil size [20] and the questionnaires of the STOP-Bang and ESS were imputed using multivariate imputation by the chained equations package (MICE, version 3.14.0) in R (R Foundation for Statistical Computing, Vienna, Austria) [21]. Baseline characteristics were analyzed using the Mann-Whitney U test for continuous variables and Fisher’s exact test for categorical variables. Spearman’s rank correlation was used to determine correlations between baseline characteristics. To examine the predictive performance of the STOP-Bang and ESS questionnaires as screening tools for OSA, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were calculated using 2×2 contingency tables. Measurements before and after surgery were examined using the Wilcoxon signed-rank test, and median differences were estimated using the Hodges-Lehmann test and presented as a 95% confidence interval (CI) [22]. The Bland and Altman method calculated the correlation coefficient of repeated measures within individuals [23]. To estimate changes in the REI across surgery, we applied a linear mixed-effects model with a random intercept for each participant, adjusting for preoperative variables selected by the least absolute shrinkage and selection operator (Lasso) method (R package: glmmLasso, version 1.6.2) [24]. The Lasso method shrinks some variable coefficients and sets others to zero, thus attempting to maintain good variables for subset selection. To evaluate the factors associated with changes in REI across surgery, we applied another linear mixed-effects model with a random intercept for each participant, adjusting for preoperative and postoperative variables selected using the Lasso method. All comparisons were two-sided, and P values <0.05 were considered statistically significant. All analyses were performed using R version 4.3.1.
RESULTS
Baseline characteristics of the study participants
Among the 42 eligible patients, four were excluded: two were being treated with continuous positive airway pressure (CPAP), and two refused to participate. Of the 38 patients enrolled, three were not included in the analysis: one did not undergo surgery because of the absence of pituitary tumors on MRI, and two had invalid respiratory polygraphy tests. Of the 35 patients with acromegaly, 24 (68.6%) had OSA (REI ≥5/hour), 15 (42.9%) had moderate-to-severe OSA (REI ≥15/hour), and 4 (11.4%) had severe OSA (REI ≥30/hour). None of the patients had central sleep apnea—defined as five or more central apneas and/or central hypopneas per hour of sleep [25]. Three patients had been diagnosed and treated for OSA (uvulopalatoplasty in two patients and use of CPAP for 1 month in the 3 years before enrolment in one patient). The remaining patients were not diagnosed with OSA. Although all study patients were naïve to endoscopic transsphenoidal surgery, three were medically treated with bromocriptine, octreotide, or cabergoline. The baseline characteristics of the 35 patients (median age, 47 years; 40% men; median BMI, 24.4 kg/m2) according to the presence of OSA and moderate-to-severe OSA are presented in Table 1 and Supplemental Table S1. Most patients with acromegaly self-reported snoring, and approximately half of them reported experiencing apnea during sleep. The median REIs were 19.6 and 3.4/hour in patients with and without OSA, respectively. The median REI and ODI values in the supine position were 29.8 and 32.2/hour, respectively, in patients with OSA. The proportion of macroadenoma (≥1 cm) was not different between patients with OSA and without OSA (91.7% vs. 81.8%, P=0.575). However, the median IGF-1 levels were higher in patients with OSA than in those without OSA (2.2×upper limit of normal [ULN] vs. 1.3×ULN, P=0.003). Correlations between sleep parameters and IGF-1 levels at baseline are shown in Fig. 1. Serum IGF-1 levels were significantly positively correlated with REI (ρ=0.53, P=0.001), ODI (ρ=0.56, P<0.001), and TS90 (ρ=0.46, P=0.006), whereas they were significantly negatively correlated with the lowest SpO2 (ρ=–0.43, P=0.009).
Screening of OSA in patients with acromegaly
The predictive performance of the STOP-Bang questionnaire is summarized in Table 2. The sensitivity, specificity, PPV, and NPV of a STOP-Bang score ≥3 were 93.3% (95% CI, 68.1% to 99.8%), 35.0% (95% CI, 15.4% to 59.2%), 51.9% (95% CI, 31.9% to 71.3%), and 87.5% (95% CI, 47.3% to 99.7%), respectively, to detect moderate-to-severe OSA. Corresponding statistics of ESS >10 were 26.7% (95% CI, 7.8% to 55.1%), 75.0% (95% CI, 50.9% to 91.3%), 44.4% (95% CI, 13.7% to 78.8%), and 57.7% (95% CI, 36.9% to 76.6%), respectively.
Effects of endoscopic transsphenoidal surgery on OSA
Biochemical remission of acromegaly was achieved in 32 (91.4%) of the 35 patients. One patient did not require a postoperative OGTT for the assessment of remission, as the pituitary gland itself was a tumor, and panhypopituitarism occurred after tumor resection. None of the patients initiated OSA treatment before the postoperative respiratory polygraphy. The median differences in REI (i.e., postoperative median REI–preoperative median REI) were –9.5/hour (95% CI, –13.3 to –5.3) in the entire cohort (Table 3, Fig. 2A) and –14.4/hour (95% CI, –19.4 to –9.9) in the 24 patients diagnosed with OSA preoperatively (Supplemental Table S2). Twelve of the 24 patients diagnosed with OSA preoperatively had an REI <5/hour postoperatively. Changes in the ODI were similar to those observed in the REI (Table 3, Fig. 2B, Supplemental Table S2). Other parameters indicating the degree of hypoxemia, such as TS90 and mean and lowest SpO2, significantly improved after surgery (Fig. 2C-E). However, changes in ESS and body weight were not significantly different across the surgeries (Fig. 2F, G). Serum IGF-1 significantly decreased after surgery (Fig. 2H). Fig. 3 shows the median differences of REI according to age (≥50 years vs. <50 years); sex; smoking status (ever smoker vs. never smoker); severity of OSA (baseline REI ≥15/hour vs. <15/hour); presence of comorbidities, such as hypertension, diabetes mellitus, and obesity (baseline BMI ≥25 kg/m2); macroadenoma (vs. microadenoma); baseline IGF-1 level (≥2×ULN vs. <2×ULN); remission of acromegaly in postoperative 3 to 6 months (vs. no remission); and weight gain (vs. stable vs. loss). The results from these subgroups were consistent with the overall results.
Using the Bland and Altman method for repeated measures within individuals, changes in weight and neck circumference did not correlate with improvements in the REI (r=0.13, P=0.436; r=0.26, P=0.122) (Fig. 4A, B). The decrease in GH and IGF-1 levels correlated with an improvement in REI (r=0.35, P=0.037; r=0.64, P<0.001) (Fig. 4C, D). Similar findings were observed in the linear mixed-effects models. REI decreased by 10.3/hour (95% CI, 13.2 to 7.5) after surgery, adjusting for age and baseline REI—those selected by the Lasso method among preoperative variables at baseline (model 1 in Table 4). Change in REI across surgery was related to baseline REI and changes in IGF-1 levels—those selected by the Lasso method among preoperative and postoperative variables (model 2 in Table 4).
DISCUSSION
Our prospective study found OSA in 68.6% of the patients with acromegaly. Serum IGF-1 levels were significantly correlated with REI at baseline. A STOP-Bang score ≥3 showed high sensitivity and NPV for the detection of moderate-to-severe OSA in patients with acromegaly. After endoscopic transsphenoidal surgery, REI decreased by approximately 10/hour, and half of the 24 patients with OSA had REI <5/hour. Improvement in REI across surgery was related to the baseline REI and change in IGF-1 levels.
Our findings are consistent with those of previous studies in that OSA is highly prevalent in patients with acromegaly, and the treatment of acromegaly reduces the prevalence and severity of OSA [4,26]. Our study showed that the median difference of REI across surgery was –9.5/hour (95% CI, –13.3 to –5.3). The degree of improvement in the REI in our study was similar to that in the AHI in a prospective Chinese study conducted using polysomnography in 24 patients with acromegaly who achieved remission after transsphenoidal surgery [27]. However, this study has several novel findings that merit further discussion. First, we showed that a STOP-Bang score ≥3 had a high sensitivity of 93.3% and an NPV of 87.5% for detecting moderate-to-severe OSA in individuals with acromegaly. Recent guidelines on the diagnosis and treatment of acromegaly comorbidities recommend careful assessment to screen for OSA at the diagnosis of acromegaly, including history, questioning of spouse/ partner, and potential use of a sleep questionnaire, such as the ESS [5,6]. However, in our study, patients with acromegaly were mostly symptomatic; over 90% had habitual snoring, and approximately half had witnessed apnea or excessive daytime sleepiness. Moreover, the predictive values of an ESS >10 were lower than those of a STOP-Bang score ≥3 in detecting moderate-to-severe OSA in patients with acromegaly. The sensitivity and NPV of an ESS >10 were 26.7% and 57.7%, respectively. The STOP-Bang questionnaire was developed as a reliable and easy-to-use screening tool in preoperative clinics [7,14] and has been validated in various populations, such as the general or obese population and bus drivers [14]. Although the recent guidelines recommend using the ESS as a sleep questionnaire to screen for OSA in patients with acromegaly [5,6], it measures the general level of daytime sleepiness across various sleep-related breathing disorders, including OSA, central sleep apnea, narcolepsy, and idiopathic hypersomnia [17]. We suggest that the STOP-Bang questionnaire, along with the ESS, can be used as screening tools for identifying moderate-to-severe OSA in patients with acromegaly. If patients with an acromegaly score of 0 to 2 on the STOP-Bang questionnaire are at low risk of OSA, the possibility of these patients having moderate-to-severe OSA can be ruled out. Otherwise, if patients with acromegaly score 3 or more on the questionnaire, a sleep study, such as polysomnography or respiratory polygraphy, should be considered.
Second, we adjusted the baseline REI, the variable selected by the Lasso method, to estimate the change in REI after surgery because the degree of change in a certain parameter would be dependent on the baseline value. As a result, REI decreased by 10.3/hour (95% CI, 13.2 to 7.5) after surgery. Third, we identified the factors associated with the effects of endoscopic transsphenoidal surgery on REI improvement. Changes in IGF-1 levels, in addition to baseline REI, were significantly related to changes in REI across surgery. Fourth, we showed postoperative improvement in various parameters related to the severity of OSA, including parameters indicating the degree of hypoxemia, such as ODI, TS90, mean and lowest SpO2, and conventional parameters measuring the number of respiratory events (i.e., AHI and REI). The AHI is most commonly used to diagnose OSA and quantify its severity. However, the frequency of respiratory events such as AHI has been limited to capturing the clinical aspects and physiological consequences of OSA; it is not strongly correlated with the symptoms of OSA, the degree of nocturnal hypoxemia, measures of quality of life, and cardiovascular mortality [28,29]. Thus, alternative parameters of disease severity have been proposed, and several measures of hypoxemia have shown a better relationship with adverse health outcomes, including composite cardiovascular outcomes and mortality, than event frequency [13,30,31]. Intermittent nocturnal hypoxemia is a pathophysiological hallmark of OSA that results in OSA-associated cardiovascular morbidity [28].
Several plausible mechanisms explain the high risk of OSA development in patients with acromegaly. Facial skeletal deformities, such as mandibular enlargement and soft tissue thickening of the pharyngeal wall and tongue, are present in acromegaly. Soft tissue thickening is caused by the accumulation of glycosaminoglycans and an increase in collagen production by connective tissue [32,33]. Tissue oedema also contributes to soft tissue thickening resulting from increased renal sodium reabsorption due to the direct stimulation of epithelial sodium channels by GH and IGF‐1 [34]. GH-/IGF-1-induced sodium retention and fluid accumulation also aggravate the upper airway obstruction caused by a rostral fluid shift from the legs to the neck when recumbent. Additionally, obesity and central or primary hypothyroidism, which are frequently associated with acromegaly, may contribute to OSA development. Lowering IGF-1 levels by acromegaly treatment may lead to a decrease in soft tissue swelling and lead to an improvement in OSA severity. Indeed, IGF-1 levels correlated with tongue volume in previous studies [26,35].
Our study has some limitations, including the small number of patients and limited statistical power, especially in the subgroup analyses. Large-scale prospective studies are required to determine whether treatment of acromegaly improves the severity of comorbid OSA across various subgroups. However, acromegaly is a rare disease, and our study had the largest sample size among prospective before–after studies evaluating the effects of surgical treatment of acromegaly on OSA. In addition, we did not measure the volume of the tongue or pharyngeal soft tissues using computed tomography or MRI, whose changes would be related to changes in GH/IGF-1 levels.
In conclusion, OSA is highly prevalent in patients with acromegaly, and IGF-1 levels are significantly correlated with OSA severity. Improvement in OSA severity after endoscopic transsphenoidal surgery is related to decreased IGF-1 levels. The STOP-Bang questionnaire can be used as a reliable screening tool for OSA in patients with acromegaly. Further large-scale prospective studies are needed to demonstrate the predictive performance of the STOP-Bang questionnaire in detecting the presence or occurrence of OSA in patients with acromegaly.
XML Download