The Influence of the Menstrual Phases on Polysomnography

Purpose: The primary objective of this study is to determine how the phases of the menstrual cycle influence the results of polysomnography (PSG). Methods: Twenty-eight adult subjects who reported regular menstrual periods, last menstrual period (LMP) within 26 days of their PSG, no exogenous hormone use, no history of polycystic ovarian syndrome, and who were scheduled for diagnostic PSG at Boston Medical satisfied inclusion criteria for the study. These subjects were divided into a Follicular Cohort (days 0-13 of the cycle) or Luteal Cohort (days 14-26 of the cycle), and a one-way analysis using a t-test was performed to test the hypothesis that the follicular phase confers protection against obstructive sleep apnea (OSA). A likelihood-ratio chi-square test was also applied to assess for a statistically significant association between menstrual stage and the presence of moderate-to-severe sleep apnea (apnea-hypopnea index (AHI) > 15/h). Thus, the statistical analysis was performed using AHI as both a continuous and a categorical outcome. Results: The mean AHI for patients in the Follicular Cohort (6.1/h) was significantly lower than the Luteal Cohort (14.3/h, p = 0.033). In the Follicular Cohort, 12% of patients had moderate to severe OSA. In the Luteal Cohort, 46% of patients had moderate to severe OSA (p = 0.045). Conclusions: Subjects undergoing PSG during the follicular phase have significantly lower AHIs than those in the luteal phase. Thus, the timing of PSG acquisition for regularly menstruating women should be considered when interpreting results.


Introduction
Female sex hormones, including estrogen and progesterone, are thought to have a protective effect against obstructive sleep apnea (OSA), which manifests as decreased rates of OSA in regularly menstruating ("premenopausal") women compared to postmenopausal women or men [1].
The physiologic rationale for the pro-respiratory function of estrogen and progesterone in OSA is multifactorial: (1) both hormones increase pharyngeal dilator muscle activity, thereby resisting a collapse of the upper airway during sleep, (2) both hormones promote hypoxic and hypercapnic ventilatory responses, (3) estrogen inhibits overexpression of hypoxia-inducible factor-1 (HIF-1), a transcription factor that is responsible for decreasing resistance to fatigability in the genioglossus muscle, and (4) progesterone acts on the nuclear progesterone receptor (nPR) transcription factor to augment the ventilatory response to hypoxia [2][3][4][5].
Although estrogen and progesterone share redundancy of function as activators of the pharyngeal dilator muscles and stimulators of the respiratory centers, prior clinical research on estrogen and progesterone therapy suggests that there are distinct differences in the therapeutic efficacy for each hormone [6][7][8][9][10]. No study of progesterone monotherapy has demonstrated a therapeutic benefit for obstructive sleep apnea (Table 1). Furthermore, there is evidence that progesterone supplementation in combination with estrogen therapy may attenuate the respiratory benefits of estrogen alone [6].  Only one study, a 62-patient prospective randomized crossover study conducted by Polo-Kantola, et al., investigated short-term unopposed estrogen replacement monotherapy for menopausal OSA patients [15]. Notably, the Polo-Kantola study demonstrated a detectable therapeutic benefit of estrogen monotherapy with decreased occurrence and frequency of sleep apnea [15]. The analogous natural state of elevated estrogen in premenopausal women occurs during the follicular phase of the menstrual cycle. Therefore, rising follicular estrogen as patients approach mid-cycle ovulation could similarly translate into reduced frequency and severity of sleep-disordered breathing on PSG. We hypothesized that premenopausal women would demonstrate lower sleep apnea severity during the follicular phase of the menstrual cycle, the time of highest unopposed estrogen.

Procedure
Permission for this study was obtained from the Boston University Institutional Review Board. Informed consent was obtained from all individual participants included in the study. This study consisted of a chart review of all female patients who underwent polysomnography at the Boston Medical Center Sleep Disorders Center between September 15, 2012 and March 15, 2013 who could recall the first day of their last menstrual period (LMP) and were not using exogenous sex hormones. In order to be included, subjects had to report an LMP of no more than 26 days prior to the study to avoid the potential confounder of early pregnancy. Subjects were excluded if they reported "irregular" menstrual periods or a history of polycystic ovarian syndrome (PCOS). All women included in the study were referred for PSG to evaluate for obstructive sleep apnea. Women referred for PSG with a multiple sleep latency test (MSLT) were excluded.
As eligible subjects were identified, their polysomnogram reports were reviewed to determine the overall apnea-hypopnea index (AHI), rapid eye movement (REM) AHI, and non-rapid eye movement (NREM) AHI. The subjects' pre-study questionnaires were reviewed to obtain additional baseline characteristics, such as body mass index, race, and recent alcohol use. Subjects were then grouped by LMP. Subjects who reported their LMP within 0-13 days from the date of their PSG were placed in the "Follicular Cohort" while subjects with an LMP 14-26 days prior to their PSG were placed in the "Luteal Cohort".
The digital PSG consisted of the simultaneous recording of electroencephalogram, electrooculogram, electromyogram, electrocardiogram, respiratory effort, thermistor respiratory flow, nasal pressure, pulse oximetry, leg movement, body position, sound, video, and positive airway pressure device. The Viasys Somnostar version 9-1b recording equipment (SensorMedics, Yorba Linda, CA) was used.
The AHI was calculated as the total number of apneas, plus hypopneas, per hour with hypopneas defined as a flow reduction with a 3% oxygen desaturation or an arousal.

Questionnaires
Prior to undergoing PSG, all patients completed a medical history questionnaire. Questions include assessments of daytime sleepiness, medical history (including menstrual history), prescription and illicit drug use, alcohol use on the day of the test, caffeine use, and typical sleep habits. Additional screening questions consisted of last reported menstrual period as well as any past medical history of irregular menses or PCOS.

Statistical analysis
AHI was evaluated both as a continuous dependent variable and as a categorical dependent variable with a moderate-to-severe OSA cutoff (AHI > 15/h). A one-way analysis was performed using a t-test to compare the mean overall AHI in the Follicular Cohort with the Luteal Cohort. This analysis was repeated for REM AHI and NREM AHI. A likelihood-ratio chi-square test was applied to assess for a statistically significant association between the stage of the menstrual cycle and AHI > 15/h.

Results
Twenty-eight subjects satisfied inclusion criteria for the study. The Follicular Cohort was comprised of 17 women and the Luteal Cohort consisted of 11 women. In the Follicular Cohort, there was one woman who self-identified as White, eight as Hispanic, six as Black, and two who did not report a race. The Luteal Cohort was comprised of two Whites, five Hispanics, one Black, one Indian, and two who did not report a race. None of the women reported drinking any alcohol on the day of the study.
The mean age for all women was 39.3 years (range: 28 -51). The mean age in the Follicular Cohort was 37.9 years (range: 28 -50) compared to 42.3 years (range: 29 -51) in the Luteal Cohort (p = 0.115) ( Table 2). Age was assessed as a potential confounding variable within the dataset and confirmed to have no statistically significant influence on the reported results (p = 0.388).
The mean overall AHI in the Follicular Cohort (6.1/h) was significantly lower than in the Luteal Cohort (14.3/h) (p = 0.033, Figure 1)

Luteal Cohort (Right).
A one-way t-test was used to evaluate whether reduced Total AHI (apnea-hypopnea index) shares a statistically significant correlation with escalating estrogen during the follicular phase of the menstrual cycle. This one-way analysis confirms a statistically significant difference between the mean Total AHIs in the follicular versus luteal groups (p=0.033). Horizontal solid bars represent the mean Total AHI within each group. The error bars and standard deviations are shown for each mean. The horizontal dashed line represents the mean Total AHI of all subjects.

Discussion
The epidemiology of obstructive sleep apnea (OSA) exhibits a clear gender disparity; the prevalence of OSA in men (3.9%) is three times the prevalence in women (1.2%) [18][19][20]. This male predominance of OSA disappears after the age of 50, at which time women show approximately equal prevalence to men [1]. These findings suggest that female sex hormones play a pivotal role in governing patient susceptibility to OSA, which is further supported by research showing significantly decreased levels of estrogen and progesterone in patients who have an AHI greater than 10/hour [21]. Despite this convincing evidence for a strong link between female sex hormones and obstructive sleep apnea, the influence of the menstrual cycle on PSG results has remained largely undefined.
Prior studies investigating the hormonal influence on OSA have shown conflicting results. For example, an 11-patient prospective crossover study by Driver, et al. concluded that the luteal phase of the menstrual cycle was associated with reduced upper airway resistance in premenopausal women [22]. A second 11-patient study by Stahl, et al. showed that none of the PSG sleep or breathing parameters were significantly affected by oscillating progesterone levels during the menstrual cycle [23]. Other research has found that neither the follicular nor the luteal phase altered the rate, duration, or extent of the desaturation associated with sleep disordered breathing during NREM sleep, but that the luteal phase was associated with marginal improvements in these parameters during REM sleep [24]. It is important to recognize, however, that these prior studies were conducted using normal subjects or heterogeneous sample populations incorporating both healthy and symptomatic subjects. None of the women included in the Driver, et al. study, for example, presented with symptoms concerning for OSA and only three women were ultimately found to have an AHI of greater than 10/hour [22]. We also had a larger sample size.
Our study has unique clinical relevance because it is comprised of undiagnosed, premenopausal women referred for diagnostic PSG due to high clinical concern for OSA. In this population of high-risk patients, obtaining accurate PSG results is particularly critical for optimizing the subsequent therapeutic intervention. Our results suggest that the menstrual phases have direct clinical consequences in the diagnosis of OSA; significantly higher overall AHI values were observed in women who underwent PSG during the luteal phase, and women in the luteal phase were nearly four times more likely to return a result of moderate to severe sleep apnea ( Figure  1). Although NREM and REM AHI values did not reach statistical significance, this was likely due to limited sample size, and it is important to note that the statistical trends observed for all measures of AHI across the Follicular and Luteal Cohorts were consistent. These findings suggest that PSG results are partially a function of the menstrual phase.
There are several study limitations. First, our sole marker of menstrual phase was patient-reported last menstrual period. Although less reliable than hormonal assays, we applied gynecologic history and patient-recalled LMP in women with menstrual regularity to predict their menstrual phase. We opted to pursue this method because it is a process that would be easy to replicate in clinical settings. Performing hormonal assays on all patients referred for diagnostic PSG, by contrast, would be impractical. Not only is patient-recalled LMP easily incorporable into the pre-PSG patient assessment, but it has also proven to be highly reliable with 81% of women correctly reporting their LMP to within two days [25]. Other study limitations include the limited number of subjects and retrospective study design.
In concordance with prior research that has shown a therapeutic benefit of short-term estrogen monotherapy, our findings suggest that the follicular phase of the menstrual cycle may induce a similarly protective effect against OSA in premenopausal women. With a nearly four-fold increase in moderate to severe OSA in the Luteal Cohort, factoring in menstrual phase could be a clinically significant improvement in obtaining a correct diagnosis and treatment plan for premenopausal women with symptoms of sleep apnea. Our findings could also account for why some women with symptoms of OSA have negative PSGs. Future research should explore the function of continuous positive airway pressure (CPAP) during the menstrual cycle as it is possible that different pressures are required based on these phases. An additional avenue for future research includes performing PSGs on the same women in both phases of the menstrual cycle to assess the magnitude of the AHI difference between phases.

Conclusions
The results of this study are the first to suggest that the follicular phase of the menstrual cycle has a protective effect against sleep-disordered breathing in a population of premenopausal women who were being referred for diagnostic PSG with high clinical suspicion for OSA. AHI values were significantly higher for patients who underwent PSG during the luteal phase, whereas patients undergoing PSG during the follicular phase demonstrated lower overall REM and NREM AHI values. These findings are clinically relevant because they suggest that premenopausal patients who undergo polysomnography during the first half of their menstrual cycles may present with lower AHI values on PSG, potentially leading to the inaccurate classification of OSA severity and increased morbidity. Consideration of the LMP in the scheduling and interpretation of PSG is encouraged to help minimize these risks.

Additional Information Disclosures
Human subjects: Boston University Institutional Review Board issued approval. Ethical Approval: All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Informed Consent: Informed consent was obtained from all individual participants included in the study. Animal subjects: This study did not involve animal subjects or tissue.