* Hypothesis 1: Reduced hypothalamic gonadotropin-releasing hormone (GnRH) outflow in older men underlies the reduced mass of luteinizing hormone (LH) pulses, and is further reduced in men with untreated OSA. The investigator aim to test hypothalamic function by examining the decrement of pulsatile LH secretion to a submaximally inhibitory dose of a selective GnRH-receptor antagonist, ganirelix, in men with untreated (i.e. active) and treated OSA. This decrement will be regressed on age, and a significant reduction in slope with active OSA (vs treated OSA) indicates disease attenuation.
* Hypothesis 2: Untreated OSA further attenuates the age-related erosion of testosterone's negative-feedback control of LH. The investigators aim to assess pituitary function by quantitating LH secretion under enforced androgen deprivation, in men with active and treated OSA. Slopes obtained by age regressions will be compared.
* Hypothesis 3: Impaired Leydig-cell steroidogenesis causes hypoandrogenemia in older men, but the effect of untreated OSA in humans is entirely unknown. Hence the investigators aim to assess testicular function by examining testosterone secretion during experimentally controlled pulsatile i.v. infusion of recombinant human LH under maximal GnRH-receptor blockade, in men with active and treated CPAP. Slopes will again be compared.
Accordingly the immediate goal is to characterize all four of hypothalamic, pituitary, testicular and feedback-dependent control of LH and testosterone secretion to determine which primary mechanism(s) occur(s) and drive(s) alterations at the other loci, in any given subject. To meet this challenge, the investigators will implement an innovative comprehensive analytical platform, that correctly embodies time-lagged, dose-dependent and nonlinear signaling among GnRH, LH and testosterone. To ensure the necessary age continuum, volunteers whose ages span 30-70 years will be study with 3-4 men per decade of life (16 men in total). To assess disease effects, the investigators will compare age regressions between men with active OSA (on sham therapy) against these same men being treated with real CPAP, and if feasible, with a convenient sample of 16 other individually age- and BMI- matched (±2.5 years and ±2.5 kg/m2) normal controls that will be selected from a larger cohort of 100 men being identified in Rochester -Minnesota (MN). The within subject cross-over design controls for possible known and unknown confounders, however BMI will be also restricted and exclude men with diabetes mellitus (by HbA1c).
Study Subject Recruitment: 16 men with OSA will be recruited aged 30-70 years, using the same entry criteria and age range utilized to generate our preliminary data 42. Additionally, body mass index will be restricted to 30-35 kg/m2. Volunteers will be recruited from clinics at University of California, Los Angeles (UCLA) -affiliated medical centers at Harbor and Santa Monica and by advertisement.
Study Design: Randomized sham-controlled cross-over trial, each period of 3 months duration and with one month washout. This design replicates a recently completed study. Another comparable study recently completed in California (and elsewhere in the USA) involved 6 months sham treatment, and many of the 1105 adults recruited had severe OSA.
Dynamic Testing of the Male Gonadal Axis: Enrolled subjects undergo repetitive sampling of peripheral blood (1.5 mL) every 10-min for 5 hr from 0800-1300 h \[2 outpatient protocols\], for 15 hr overnight (2200 - 1300h) and for 22 hr overnight (1500 - 1300h). Polysomnography (2200-0600h) is performed during the 2 inpatient visits. A terminal i.v. bolus injection of GnRH (100 ng/kg) is given at 1100h during all 4 visits to calibrate endogenous GnRH response. Ganirelix (or saline) is given at 2000h or 1500h: hypothalamic function is tested by partial (submaximal) ganirelix blockade of endogenous GnRH. Oral medications (Δ) are placebo except during pituitary testing when response of LH to androgen depletion is enforced by ketoconazole (KTCZ) with adrenal rescue with dexamethasone (DEX). Here, KTCZ (1000mg) and DEX (0.75mg) are given at 2000h, KTCZ (400mg) at 0800h and DEX (0.75mg) at 1300h. Testis function is assessed by testosterone response to a fixed exogenous LH stimulus: 6 pulses of rh LH (18.75 IU each) every 2 hrs. from 2300h under maximal ganirelix blockade of endogenous LH.
Statistical Analysis: The effect of age and disease on each mechanistic outcome will be assessed by constructing a repeated measures mixed model which includes OSA (treated versus untreated), age and the OSA\*age interaction. The statistical significance of the interaction term indicates whether the disease process (OSA) alters the age related decline in these outcomes (i.e. assesses difference in the slopes under the two conditions). The within-subject design should reduce variance and increase power. Assuming that a suitable convenient control group can be assembled in Rochester MN, the investigators will additionally construct between-subject mixed models where the disease effect will be untreated OSA versus normal controls, and the treatment effect will compare treated OSA versus normal controls. The former will affirm disease effects and the latter whether treatment for 3 months with CPAP fully reverses deficits to normality.
Power Analysis: Preliminary data suggest that untreated OSA attenuates testosterone feedback, resulting in impaired unleashing of LH pulse. Under ketoconazole castration, the correlation between LH pulse frequency and age in normal men is -0.50. A sample size of 16 men has 80% power with two-tailed á=0.05 to detect an attenuation of this correlation by 1.00. By way of comparison, the investigators observed a one-third greater attenuation of correlation of 1.35. Hence, power is adequate. This estimate is conservative since the within-subject study design, which reduces variability, was not incorporated and the ketoconazole-induced testosterone depletion paradigm to assess pituitary function should further sharpen age regressions as the variability of each individual's blood testosterone concentration is reduced. The latter has been verified in normal men, and should also be true in men with OSA.
Outcomes: To appraise GnRH outflow (Aim 1), the response variable is the ganirelix-suppressed (analytically reconstructed) basal LH secretion. To assess gonadotrope responsiveness, the pertinent end point is the mass of LH secreted following the exogenous GnRH stimulus. The degree of feedback unleashing is inferred by the degree of elevation of LH pulse frequency elicited by androgen withdrawal (Aim 2). And, Leydig-cell sensitivity is defined by analytical reconstruction of rate of testosterone secretion achieved in response to the last two of six (pseudosteady-state) pulses of recombinant human LH (Aim 3).
Other Outcomes: GnRH outflow will be reconstructed and simultaneously calculate hypothalamic and pituitary testosterone feedback, as previously described. These data can be analyzed by mixed model as described in statistical analysis (before). Pulsatile LH secretion aligns with slow wave sleep at the onset of puberty, implicating sleep architecture in the timing of puberty. The investigators will perform the same analysis described, by comparing (by Student's t-test) the proportion of LH pulses that occur during slow wave versus non-slow wave sleep, normalized for the amount of time spent in each sleep stage. The final additional outcome will be to examine total testosterone secretion overnight during the control condition in the first phase of the study, separately and after combining with findings from the original 18 men described in preliminary data.
Anticipated and Alternate Results: It is expected to be showed that OSA impairs testosterone feedback (i.e. pituitary dysfunction), specifically that androgen depletion (Aim 2), will increase LH pulse frequency and reduce basal LH secretion. Although It is postulated that reduced basal LH secretion is due to attenuated testosterone feedback at the pituitary, an unexpected but plausible alternate finding would be that altered GnRH outflow is responsible - which It is assessed in Aim 1 by graded GnRH receptor blockade. Additionally, due to recent analytical innovations, the investigators can also reconstruct hypothalamic versus pituitary feedback effects of testosterone (Aim 2) whilst simultaneously assessing GnRH outflow (Aim 1). This is particularly pertinent since preliminary data of brain permeant versus brain impermeant androgen receptor blockade suggests specific brain effects. Specific brain effects are plausible because the intermittent hypoxia of OSA is already known to alter neurocognition. Another inexplicable finding is that the relative reduction in LH with OSA seems to be greater than the corresponding fall in testosterone, which might suggest relatively enhanced Leydig cell responsiveness. Indeed, intermittent hypoxia increases testosterone production from primary rat Leydig cell cultures in response to human chorionic gonadotropin (hCG), but the effect of OSA on testicular steroidogenesis in humans is not known. Aim 3 will address this possibility for the first time using a physiological rh LH probe. Testosterone secretion is expected to increase with CPAP. If it does not, in the presence of preserved or enhanced Leydig cell responsiveness to rh LH (Aim 3), then this would motivate a future longer duration study of CPAP.
Potential Problems and Alternate Strategies: the investigators do not anticipate recruitment problems since our study requires enrolling a single individual each month for 16 months, however recruitment could be broadened to the Westwood and other UCLA-affiliated clinics, if needed. If the cross-over design proves problematic, then the investigators can convert to a 3 month parallel group study of 32 men without loss of power. The study includes a run-in period wherein CPAP compliance will be monitored by direct CPAP machine download and subjects excluded if non-compliant, since intensive endocrine testing is not performed until after 3 months of established therapy. Our study is not dependent upon being able to include a convenient control group where each control is within ±2.5 years and ±2.5 kg/m2 of each case. Finding 16 controls from the 100 men aged 18-80 years, with BMI of 18-35 kg/m2 and without significant illnesses (including diabetes mellitus) being identified should be possible. If not, the investigators will loosen the matching criteria (by doubling intervals to ±5.0 years or kg/m2), and simultaneously match 2 controls for every case. A further problem might be if LH and testosterone assays are not aligned, however the investigators will test this directly by assaying a single pooled sample from each individual (16 men in total) for testosterone by tandem mass spectroscopy and for LH by immunoassay and against external standards to ensure comparability.
