The effects of testosterone on ventilatory responses in men with obstructive sleep apnea: a randomised, placebo-controlled trial

Transcription

1 J Sleep Res. (213) 22, Obstructive sleep apnea The effects of testosterone on ventilatory responses in men with obstructive sleep apnea: a randomised, placebo-controlled trial ROO KILLICK 1,2, DAVID WANG 1,2, CAMILLA M. HOYOS 1, BRENDON J.YEE 1,2, RONALD R. GRUNSTEIN 1,2 and P E T E R Y. L I U 1,3 1 NHMRC Centre for Integrated Research and Understanding of Sleep (CIRUS), Woolcock Institute of Medical Research, University of Sydney, Sydney, NSW, Australia, 2 Department of Respiratory & Sleep Medicine, Royal Prince Alfred Hospital, Sydney, NSW, Australia and 3 David Geffen School of Medicine at UCLA, Harbor-UCLA Medical Center and Los Angeles Biomedical Research Institute, Torrance, CA, USA Keywords obstructive sleep apnea, sleep-disordered breathing, testosterone, ventilatory chemoreflexes, ventilatory control Correspondence Peter Y. Liu, Endocrine and Cardiometabolic Research, Woolcock Institute of Medical Research, University of Sydney, Glebe, NSW 237, Australia. Tel.: ; fax: ; pliu@mail.usyd.edu.au Accepted in revised form 25 November 212; received 17 July 212 DOI:.1111/jsr.1227 SUMMARY We recently showed that testosterone therapy worsens sleep-disordered breathing at 6 7 weeks, but not after 18 weeks, in men with obstructive sleep apnea. Changes in ventilatory chemoreflexes may be responsible. The effect of testosterone on ventilatory chemoreflexes in men with obstructive sleep apnea has not been systematically studied before. Twenty-one obese men with obstructive sleep apnea, a subgroup of our recent report, were randomised in an 18-week, randomised, doubleblind, placebo-controlled, parallel group trial to three intramuscular injections (, 6, 12 weeks) of either mg testosterone undecanoate (n = ) or placebo (n = 11). Awake ventilatory chemoreflex testing was performed before (week ), during (week 6) and at the end of treatment (week 18) to determine the ventilatory carbon dioxide recruitment threshold and chemosensitivity. Sleep and breathing was assessed by overnight polysomnography at, 7 and 18 weeks. Serum hormones levels were measured at every visit. A significant increase in blood testosterone levels (5.65 nmol L 1,.51.8 nmol L 1, P =.3) and lean muscle mass (2.36 kg, kg, P =.7) between the two groups was observed as expected. No significant differences were seen in ventilatory chemoreflexes between the two groups at 6 weeks or at 18 weeks. However, positive correlations were observed between changes in serum testosterone and hyperoxic ventilatory recruitment threshold (r =.55, P =.3), and between changes in hyperoxic ventilatory recruitment threshold and time spent with oxygen saturations during sleep <9% (r =.57, P =.3) at 6 7 weeks, but not at 18 weeks. Time-dependent alterations in ventilatory recruitment threshold may therefore mediate the time-dependent changes in sleep breathing observed with testosterone. INTRODUCTION Obstructive sleep apnea (OSA) is characterised by recurrent episodes of upper airway occlusion during sleep, leading to transient alterations in gas exchange and increasing inspiratory effort against the occluded airway. OSA syndrome affects 9% of middle-aged men and 4% of women (Young et al., 1993), and is associated with increased mortality and many co-morbidities. OSA is 3.3-fold more prevalent in men than women, and 4.5-fold more prevalent in postmenopausal women than premenopausal women (Bixler et al., 21). Although anatomical differences in the upper airway likely contribute to these gender differences, several studies have postulated that these gender imbalances in sleep-disordered breathing (SDB) could be secondary to differences in ventilatory control due to sex hormones, such as testosterone (Ahuja et al., 27; White et al., 1985; Zhou et al., 23). The foundation for this hypothesis is the recognition that altered ventilatory control may cause SDB (Younes, 28). However, important limitations in the available literature 331

2 332 R. Killick et al. include the lack of placebo controls in studies and the extrapolation of findings in healthy normals to patients with sleep apnea. The latter is especially problematic as ventilatory responses differ between those with and those without OSA (Gold et al., 1993). Studies examining the effect of testosterone on ventilatory responses in a sleep apneic population are therefore required. Concurrently, there is growing interest in the use of androgen supplementation in obese and older men to improve body composition, muscle and bone strength, and physical functioning, and this affords the opportunity to assess sleep, breathing and ventilatory effects in this patient group. In 23, we showed that short-term, high-dose testosterone reduced sleep and worsened breathing in older healthy men. These effects occurred with just 3 weeks of treatment, which rendered anatomical changes unlikely and, in any case, upper airway narrowing was not detected (Liu et al., 23). Due to the rapid time course, we proposed that the effects of testosterone on SDB might be mediated through changes in ventilatory chemoreflex mechanisms, although these were not assessed in that study. More recently, we conducted a randomised, double-blind, placebo-controlled, parallel group study to investigate the time-dependent effects of testosterone on sleep and breathing in 67 obese men with severe OSA (Hoyos et al., 212). A worsening of the percentage of total sleep time spent below 9% oxygen saturations (SpO 2 T9%), a key marker of overnight hypoxemia, was seen at 6 7 weeks, but not at 18 weeks. This was accompanied by similar time-dependent trends in the apnea hypopnea index (AHI). To explore the underlying mechanisms that might mediate these time-dependent relationships between testosterone and OSA, we also evaluated ventilatory chemoreflexes on a subset of 21 of the original 67 subjects in that study. In the sub-study, we hypothesise that there is a time-dependent association between testosterone serum concentration, ventilatory chemoreflexes and the degree of SDB. We now report for the first time the findings from the sub-study, which is the first double-blind, placebo-controlled clinical trial investigating the effect of testosterone on ventilatory chemoreflexes and breathing during sleep in patients with OSA. MATERIALS AND METHODS The study was sponsored by the Sydney South West Area Health Service, and complied with Good Clinical Practice guidelines, applicable regulatory requirements and the Declaration of Helsinki. All subjects provided written informed consent, and both the patient information sheet and study protocol were approved by the Sydney South West Area Health Service Human Research and Ethics Committee (RPAH Zone). The study is registered with the Australia New Zealand Clinical Trials Network, number ACT- RN This was a randomised, double-blind, placebo-controlled, parallel group study. It was a subset of the last 21 consecutive men enrolled from a larger study who were additionally required to undergo ventilatory response testing. The full description of the full study, minus the ventilatory chemoreflex tests described here, is reported elsewhere (Hoyos et al., 212). In brief, obese, adult men with OSA were randomly allocated to receive three intramuscular injections of either Reandron e (testosterone undecanoate mg in 4 ml castor oil vehicle; Bayer Schering Pharma AG, Berlin, Germany) or 4 ml oil vehicle placebo at, 6 and 12 weeks. Ventilatory chemoreflex testing was performed at baseline and immediately before the injection at weeks 6 and 18. Other relevant secondary outcomes measured included: sleep and breathing variables by in-laboratory polysomnography (PSG; Sandman Elite v9.2, Tyco Healthcare, Denver, CO, USA) at, 7 (about days after the second injection) and 18 weeks; blood hormone samples for luteinising hormone (LH), folliclestimulating hormone (FSH) and testosterone concentrations at, 6, 7, 12 and 18 weeks; and anthropometric data. Blood hormones and PSG were measured by standard techniques in these men as we have previously reported (Hoyos et al., 212). All researchers were blinded to the treatment allocation for the duration of the study. Ventilatory chemoreflex tests Ventilatory chemoreflex testing was performed using Duffin s modified rebreathing method, which is described in full elsewhere (Duffin et al., 2; Wang et al., 211). Briefly, subjects performed two iso-oxic rebreathing trials, involving 5 min of voluntary hyperventilation of room air (end-tidal pco 2 maintained between 19 and 25 mmhg), then 5 min of rebreathing through a closed circuit, during which time the subjects breath-by-breath responses were analysed in terms of minute ventilation (VE) against end-tidal pco 2. The first test was performed under iso-hyperoxia (po 2 15 mmhg) to determine the ventilatory response to hypercapnia. The second test was performed after a 45-min resting interval (po 2 5 mmhg), measuring the response to hypercapnia under iso-hypoxia. Continuous real-time gas sampling was obtained with a CO 2 /O 2 analyser, and the sampled gas was returned to the circuit. Iso-oxia was maintained by adjustment of entrained oxygen, by continuous computerised analysis of expired gas concentrations. The rebreathing container was situated within a rolling seal spirometer, and measurements of minute ventilation were made and stored electronically. Rebreathing was continued for 5 min or until a clear physiological increase in VE was seen, whichever was shorter. All subjects performed the tests in the morning, in the fasting state, in a quiet room, without any external stimulation, supervised by the same two researchers. Results were automatically stored and analyses performed using software provided by Prof. James Duffin from Toronto University (National Instruments, LabVIEW, Austin, TX, USA). The basal VE, carbon dioxide ventilatory recruitment threshold (VRT CO2 ) and chemosensitivity of ventilation to increases in CO 2 were quantified where possible and cross-

3 Effects of testosterone on ventilatory responses 333 checked by two researchers who are blinded to the treatment groups. Hyperoxic/hypoxic chemosensitivity was calculated as the slope of VE against end-tidal pco 2. Peripheral chemosensitivity was calculated by subtracting hyperoxic chemosensitivity from hypoxic chemosensitivity (Wang et al., 211). Some responses could not be analysed and were excluded if a clear physiological response was not achieved as determined by the assessors. Statistical analysis Analyses were performed using SAS version 9.2 (SAS Institute, Cary, NC, USA). Outcome variables were the calculated differences from baseline at 6 and 18 weeks. Overall and individual time point differences between groups were assessed using mixed model analyses of treatment, time and the interaction. Relationships between ventilatory chemoreflexes, sleep breathing and testosterone levels were explored using Pearson s correlation coefficient. Data were considered significantly different at P <.5 (two sided). RESULTS Twenty-one subjects were randomised to receive either testosterone (n = ) or matching placebo (n = 11). Six subjects withdrew (three from each group), with 18 completing testing at week 6 and 15 completing week 18. Withdrawals were due to personal reasons or time commitments (n = 5), and one subject withdrew due to finding the injections painful. There were no significant differences between groups at baseline in this subset (Table 1). Subjects were men (mean age 47.6 years), obese (mean body mass index 37.7 kg m 2 ), with severe OSA (mean AHI 32.6 events h 1 ), and had significant overnight hypoxia (mean ODI 28.8 per h) assessed by PSG. From the mixed model analyses (Table 2), testosterone therapy suppressed gonadotrophins (LH P =.8, FSH P <.1), and increased testosterone concentrations (P =.3) and lean muscle mass (P =.7) between groups, as expected. No statistically significant mean difference was found in any key ventilatory chemoreflex or PSG variable following testosterone treatment overall, at 6 weeks or at 18 weeks (Table 2). A significant positive linear relationship was seen between change in hyperoxic VRT CO2 and change in serum testosterone levels between week 6 and baseline (r =.55, P =.3; Fig. 1a). Additionally the change in hyperoxic VRT CO2 at week 6 correlated to the change in SpO 2 T9% at the PSG performed at week 7 (r =.57, P =.3; Fig. 1b). These correlations did not, however, persist at week 18 (Fig. 1c,d). A trend between changes in hypoxic VRT CO2 and serum testosterone at week 6 was seen (r =.59, P =.6, not shown), but once again did not persist to week 18. DISCUSSION We observed for the first time significant time-dependent positive correlations between the change in serum testosterone and the change in VRT CO2, along with a relationship between VRT CO2 and hypoxia during sleep in OSA subjects. However, differences in ventilatory control between the testosterone and placebo groups were not seen. Nevertheless, these findings suggest that altered ventilatory responses may be a key mechanism of increased SDB with short-term testosterone. Larger studies will be required to investigate this more definitely. At present there is limited literature looking specifically at the effects of testosterone on ventilatory responses (Ahuja et al., 27; Mateika et al., 24; Matsumoto et al., 1985; Schneider et al., 1986; White et al., 1985), and these have been performed only in subjects without documented SDB, using varying methodologies and without a unifying conclusion of the effect of testosterone. However, the available literature does provide a mechanism by which alterations in VRT CO2 by testosterone in men with OSA, as we propose, could worsen SDB. Previous data showed that suppressing testosterone in 11 healthy men with leuprolide acetate decreased VRT CO2 during wakefulness in both hyperoxic and hypoxic conditions, and in a subset of five men during non-rapid eye movement (NREM) sleep (Mateika et al., 24). Additionally, when transdermal testosterone was administered to eight healthy, pre-menopausal females, an increase in both apneic threshold and hypocapnic ventilatory response during NREM sleep was seen (Zhou et al., 23). These two studies support our findings that suggest that testosterone may decrease breathing stability predominantly through an increase in VRT CO2. Indeed, a shift in VRT CO2 closer to the resting level may indicate a narrower CO 2 reserve and therefore a greater chance of SDB (Dempsey, 25). Besides the effects on VRT CO2, differing effects of testosterone on both the hypoxic and hypercapnic chemoresponses have been described (Mateika et al., 24; Matsumoto et al., 1985; Schneider et al., 1986; White et al., 1985). White et al. (1985) administered testosterone to 12 hypogonadal men, and showed an increase in hypoxic ventilatory responses and metabolic rate, but not hypercapnic responses. Interestingly, there were no correlations seen between serum testosterone level, nor change in testosterone level, and ventilatory parameters, unlike our observations. In other work, ventilatory chemoreflex tests were performed on 11 healthy women treated with testosterone (Ahuja et al., 27). Testosterone increased resting VE as well as ventilatory chemosensitivity under hyperoxia. We speculate that the relationship we have seen between hyperoxic VRT CO2 and testosterone level and then with SpO 2 T9 may suggest that testosterone could produce a direct central effect on ventilatory chemoresponsiveness within the brainstem, as the correlation may suggest a role of the central chemoreceptors rather than the peripheral

4 334 R. Killick et al. Table 1 Baseline participant characteristics (mean SEM) Testosterone (n = ) Placebo (n = 11) P-value Age (years) BMI (kg m 2 ) Weight (kg) Waist circumference (cm) Lean muscle (kg) Fat (kg) Hyperoxic VRT CO2 (mmhg) Hyperoxic chemosensitivity (L min 1 mmhg 1 ) Hyperoxic basal ventilation (L min 1 ) Hypoxic VRT CO2 (mmhg) Hypoxic chemosensitivity (L min 1 mmhg 1 ) Hypoxic basal ventilation (L min 1 ) Peripheral chemosensitivity-calculated (L min 1 mmhg 1 ) Serum testosterone (nmol L 1 ) AHI (events h 1 ) ODI 3% (events h 1 ) SpO 2 T9% (%total TST) SpO 2 nadir (%) AHI, apnea hypopnea index; BMI, body mass index; ODI 3%, oxygen desaturation index; SpO 2 T9%,% time of total sleep time (TST) arterial oxygen saturation <9%; VRT CO2, carbon dioxide ventilatory recruitment threshold. Table 2 Mixed model outcomes Mean change from baseline Testosterone Placebo Differences of means (95% CI) P-value BMI (kg m 2 ) (.55 to 1.44).36 Weight (kg) ( 1.82 to 4.29).41 Waist circumference (cm) ( 2.88 to 4.98).58 Lean muscle (kg) (.8 to 3.93).7* Fat (kg) ( 1.79 to 4.9).41 Hyperoxic VRT CO2 (mmhg) ( 3.16 to 5.93).53 Hyperoxic chemosensitivity (L min 1 mmhg 1 ) ( 1.11 to.67).61 Hyperoxic basal ventilation (L min 1 ) ( 1.28 to 5.2).23 Hypoxic VRT CO2 (mmhg) ( 5.16 to 6.96).75 Hypoxic chemosensitivity (L min 1 mmhg 1 ) ( 3.2 to 1.74).53 Hypoxic basal ventilation (per min) ( 4.94 to 5.78).87 Peripheral chemosensitivity-calculated (L min 1 mmhg 1 ) ( 3.15 to 3.7).98 Serum testosterone (nmol L 1 ) (.51 to.8).3* LH (U L 1 ) ( 1.24 to 3.87).8* FSH (U L 1 ) ( 2.15 to 5.4) <.1* Prostate specific antigen (ng ml 1 ) (.17 to.47).3* BMI, body mass index; FSH, follicle-stimulating hormone; LH, luteinising hormone; VRT CO2, carbon dioxide ventilatory recruitment threshold. *denotes significance of P <.5. chemoreceptors, therefore worsening SDB in the short term but not the longer term. This therefore provides a mechanism by which testosterone s effect on SDB may be time dependent. The exact molecular, neuronal or other changes that occur to enforce these time dependencies will require future evaluation. A limitation in our study was the relatively small number of participants in the subset, and indeed reliable data were not obtained for all variables at each visit, although the sample size is already larger than most of the studies in the field. Secondly, testing ventilatory chemoreflexes during sleep would have been useful to validate our findings during wake and provide more insight into the pathogenesis of SDB. Thirdly, this study did not measure changes in upper airway anatomy, although our previous work (Liu et al., 23) did not

5 Effects of testosterone on ventilatory responses 335 Figure 1. Correlations between changes in ventilatory response variables and changes in sleep parameters. Data are scatterplots of Pearson s correlations between changes from baseline in (a) hyperoxic ventilatory recruitment threshold to carbon dioxide (VRT) and serum testosterone at week 6 (n = 15), (b)% time of total sleep time (TST) that the arterial oxygen saturation was <9% (SpO 2 T9%) and hyperoxic VRT at week 6 (n = 14), (c) hyperoxic VRT and serum testosterone at week 18 and (d) SpO 2 T9% and hyperoxic VRT at week 18 (both n = 11). Pearson s correlation (r) and P-values are indicated for each correlation. (a) Week (c) Week r =.55, P =.3 (b) Week 6 r =.57, P =.3 3 ΔSpO 2 T9 (%TST) ΔSerum Testosterone (nmol L 1 ) r =.17, P = ns (d) Week 18 r =.3, P = ns 3 ΔSpO 2 T9 (%TST) ΔSerum Testosterone (nmol L 1 ) unveil anatomical changes. Lastly, our subjects were not specifically selected on the basis of being hypogonadal, and hence different findings may apply in the population of hypogonadal men. In conclusion, we have described a time-dependent, positive linear relationship between changes in serum testosterone and VRT CO2, which then had a correlation with overnight hypoxia. Changes in VRT CO2 due to testosterone may be responsible for the short-term deterioration of SDB, which resolves with time. Establishing this definitively will require longer and larger studies. ACKNOWLEDGEMENTS We would like to acknowledge Professor James Duffin for his invaluable advice on the ventilatory response testing methodology and analysis. We thank the men who participated in the study, the sleep physicians, and those that assisted in study assessments and coordination. We also thank Dr Farid Saad for assistance in obtaining drug and placebo. This article was supported by the National Health and Medical Research Council of Australia (NHMRC) through a project grant (512499), a Centre for Clinical Research Excellence in Interdisciplinary Sleep Health (571421) and fellowships to RK, DW, CH, RRG and PYL (633161, , 51257, and 25248, respectively). Bayer Schering supplied study drug, matching placebo and $2, to date. CONFLICT OF INTEREST No author has any conflict of interest to declare. REFERENCES Ahuja, D., Mateika, J. H., Diamond, M. P. and Safwan Badr, M. Ventilatory sensitivity to carbon dioxide before and after episodic hypoxia in women treated with testosterone. J. Appl. Physiol., 27, 2: Bixler, E. O., Vgontzas, A. N., Lin, H. M. et al. Prevalence of sleepdisordered breathing in women: effects of gender. Am. J. Respir. Crit. Care Med., 21, 163: Dempsey, J. A. Crossing the apnoeic threshold: causes and consequences. Exp. Physiol., 25, 9: Duffin, J., Mohan, R. M., Vasiliou, P., Stephenson, R. and Mahamed, S. A model of the chemoreflex control of breathing in humans: model parameters measurement. Respir. Physiol., 2, 12: Gold, A. R., Schwartz, A. R., Wise, R. A. and Smith, P. L. Pulmonary function and respiratory chemosensitivity in moderately obese patients with sleep apnea. Chest, 1993, 3: Hoyos, C. M., Killick, R., Yee, B. J., Grunstein, R. R. and Liu, P. Y. Effects of testosterone therapy on sleep and breathing in obese men with severe obstructive sleep apnoea: a randomized placebo-controlled trial. Clin. Endocrinol. (Oxf.), 212, 77: Liu, P. Y., Yee, B., Wishart, S. M. et al. The short-term effects of highdose testosterone on sleep, breathing, and function in older men. J. Clin. Endocrinol. Metab., 23, 88: Mateika, J. H., Omran, Q., Rowley, J. A., Zhou, X. S., Diamond, M. P. and Badr, M. S. Treatment with leuprolide acetate decreases the threshold of the ventilatory response to carbon dioxide in healthy males. J. Physiol., 24, 561: Matsumoto, A. M., Sandblom, R. E., Schoene, R. B. et al. Testosterone replacement in hypogonadal men: effects on obstructive sleep apnoea, respiratory drives, and sleep. Clin. Endocrinol. (Oxf.), 1985, 22: Schneider, B. K., Pickett, C. K., Zwillich, C. W. et al. Influence of testosterone on breathing during sleep. J. Appl. Physiol., 1986, 61:

6 336 R. Killick et al. Wang, D., Marshall, N. S., Duffin, J. et al. Phenotyping interindividual variability in obstructive sleep apnoea response to temazepam using ventilatory chemoreflexes during wakefulness. J. Sleep Res., 211, 2: White, D. P., Schneider, B. K., Santen, R. J. et al. Influence of testosterone on ventilation and chemosensitivity in male subjects. J. Appl. Physiol., 1985, 59: Younes, M. Role of respiratory control mechanisms in the pathogenesis of obstructive sleep disorders. J. Appl. Physiol., 28, 5: Young, T., Palta, M., Dempsey, J., Skatrud, J., Weber, S. and Badr, S. The occurrence of sleep-disordered breathing among middle-aged adults. N. Engl. J. Med., 1993, 328: Zhou, X. S., Rowley, J. A., Demirovic, F., Diamond, M. P. and Badr, M. S. Effect of testosterone on the apneic threshold in women during NREM sleep. J. Appl. Physiol., 23, 94: 1 7.