INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Respiratory failure is inevitable in the course of Duchenne muscular dystrophy (DMD), although the time course of progression to it varies between individuals. The development of respiratory failure during wakefulness is usually preceded by hypoventilation during sleep (1), which appears to be a consequence of the combined effects of sleep-induced decreased ventilatory drive and respiratory muscle weakness. Although often nonobstructive in origin because of weakness of the respiratory pump muscles, the hypoventilation may have an obstructive component because of upper airway muscle weakness (4, 5), particularly when other predisposing factors, such as obesity, are present.

Potential consequences of sleep-related hypoventilation include sleep disruption with diurnal lethargy, hypoxemia, and hypercarbia. Untreated, sleep-related hypoxemia and hypercarbia may have a causal role in the development of respiratory failure because of disordered ventilatory control resulting from adaptation and downregulation of the ventilatory responses to these stimuli (6). Previous uncontrolled studies suggest that treatment of sleep hypoventilation with noninvasive ventilation (NIV) can improve daytime gas exchange (7- 9) and prolong survival (10, 11) of patients with established respiratory failure.

Hence it appears possible that sleep-related hypoventilation could both precede and contribute to the development of respiratory failure. If so, then the degree of sleep hypoventilation would be an important marker of progression of the disease and a guide to intervention with NIV during sleep. However, it is unclear when first to investigate for the possibility of sleep-related hypoventilation. Polysomnography is an expensive mode of investigation that presents logistic difficulties and inconvenience to patients with DMD. If the magnitude of sleep-related hypoventilation is related to progression of the disease, then it might be expected to relate to daytime measures reflecting severity of respiratory muscle weakness. Identification of such daytime predictors of the severity of sleep hypoventilation could permit selection of those patients in whom the possibility of hypoventilation was sufficiently high to justify polysomnography. In previous studies relating respiratory function to sleep hypoxemia, Smith and coworkers (2) categorized 14 patients with DMD as "desaturators" and "nondesaturators" and found only maximum expiratory pressure to be different between the groups, although they noted a tendency toward lower vital capacity and maximum inspiratory pressure among the desaturators. In another study of six patients, Barbe and coworkers (3) found relationships only between the apnea-hypopnea index (AHI) and diurnal hypoxemia and FEV₁/FVC.

Because of the limited number of patients and measures of sleep hypoventilation in these studies and our desire to develop objective criteria for scheduling polysomnography, we investigated the relationship between daytime measures of respiratory function and ventilation during sleep in a larger group of patients with DMD. To define further the nature of this relationship we also evaluated the effect of therapy on daytime respiratory function in those patients in whom sleep hypoventilation was identified and treated with NIV.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

All patients  12 yr old with DMD in Western Australia were identified from a central registry compiled from the Department of Neuropathology, Royal Perth Hospital (Perth, Australia). Patients were excluded if they previously had polysomnography or NIV. Of the 39 potential subjects identified, 11 had previously been assessed by polysomnography. The remaining 28 subjects were contacted by letter and telephone and asked to enroll in the study. The protocol was approved by the local ethics committee and informed consent was obtained from all subjects.

Ventilatory Capacity

Spirometry (FEV₁, FVC, and FEV₁/FVC) was performed in the sitting position according to the standards of the American Thoracic Society (ATS) (12). Total lung capacity was measured by the helium dilution technique. Maximal inspiratory and expiratory mouth pressures were measured according to the methods of Black and Hyatt (13).

Gas Exchange

Arterial puncture of the radial artery was performed in the sitting position at a standard time of day (midafternoon) after local anesthesia. The sample was analyzed (ABL520; Radiometer, Copenhagen, Denmark) for pH, base excess, and arterial oxygen and carbon dioxide partial pressures.

Polysomnography

Polysomnography was performed according to the standards of the ATS (14). Sleep staging (by electroencephologram [EEG], electrooculogram [EOG], and electromyogram [EMG]), oronasal flow (thermistor), respiratory inductance plethysmography (Respitrace, Ardsley, NY), oxygen saturation (Biox 3700; Ohmeda, Louisville, CO), and transcutaneous CO₂ (TCM3; Radiometer) were recorded onto a computerized workstation (Nicolet Instrument, Madison, WI). An apnea was defined as a cessation of airflow with a duration of at least 10 s. An obstructive apnea was defined as an apnea in which there was evidence of persisting respiratory effort. A hypopnea was defined as a reduction in airflow or chest wall motion resulting in arterial oxygen desaturation of at least 3%. The proportion of sleep time spent below a saturation of 90% (total sleep time [TST] < 90%) was derived from a cumulative frequency curve of arterial saturation and used as an indicator of sleep hypoventilation, as supplemental oxygen was not used in any subject. A threshold TST < 90% of  2% was arbitrarily chosen for analysis as an indicator of clinically significant sleep hypoventilation.

Statistical Analysis

Analysis was performed with the SigmaStat software package (Jandel, San Rafael, CA). Interrelationships between ventilatory capacity, gas exchange, and polysomnographic parameters were examined by simple linear regression analysis. The relationships of parameters of ventilatory capacity and gas exchange to each other and to sleep hypoventilation (TST < 90%) were examined by forward stepwise linear regression. Gas exchange and spirometry at baseline and after NIV in those subjects requiring therapy were compared by a paired t test. All results are presented as means ± standard deviation. p < 0.05 was considered significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Of the 28 subjects approached to participate in the study, 7 declined and 2 required intervention with NIV for hypercapneic respiratory failure before the study commenced. The remaining 19 subjects were entered into the study. All were male (in view of the genetic basis of the condition) with a mean (± SD) age of 18.6 ± 3.9 yr. Their mean body mass index (BMI) was 20.5 ± 5.2 kg/m². Lung function, gas exchange, and polysomnography are summarized in Table 1. Eleven subjects were found to have a TST < 90% of  2%, 4 of whom had more than four obstructive apneas per hour of sleep. These four subjects are separately indicated in Figures 1-3. Their mean BMI (21.5 ± 3.4 kg/m²) was similar to the group as a whole. All other subjects had less than one obstructive apnea per hour of sleep. There was no significant relationship between age and degree of desaturation (mean Sa_O2, nadir Sa_O2, or TST < 90%).

View larger version (10K): [in this window] [in a new window]   Figure 1.   Relationship between wakeful PaCO2 and FEV1. Patients with obstructive apneas (more than four per hour) are indicated by triangles. All other subjects (circles) had less than one obstructive apnea per hour.

View larger version (10K): [in this window] [in a new window]   Figure 2.   Relationship between proportion of total sleep time spent at an SaO2 of < 90% (TST < 90%) and FEV1. Patients with obstructive apneas (more than four per hour) are indicated by triangles. All other subjects (circles) had less than obstructive apnea per hour.

View larger version (14K): [in this window] [in a new window]   Figure 3.   Relationships between proportion of total sleep time spent at an SaO2 of < 90% (TST < 90%) and PaCO2 (top panel ) and base excess (bottom panel ). Patients with obstructive apneas (more than four per hour) are indicated by triangles. All other subjects (circles) had less than one obstructive apnea per hour.

Concordant decreases in Sa_O2 and increases in transcutaneous PCO₂ (Ptc_CO2) were noted particularly during rapid eye movement (REM) sleep-related hypoventilation, where invariably the lowest Sa_O2 values were observed, in association with the highest Ptc_CO2 levels. The proportion of sleep spent in REM was similar between those subjects with significant sleep hypoventilation (11.3 ± 8.7%) and those without (14.6 ± 6.5%). There were no systematic differences in sleep posture between these groups (80.1 ± 34.3 versus 65.8 ± 41% of sleep spent supine, respectively).

Relationship between Ventilatory Capacity and Gas Exchange

FEV₁ and FVC were significantly related to Pa_CO2 (r = 0.70 and 0.74, respectively, p < 0.001), Pa_O2 (r = 0.65 and 0.67, respectively, p < 0.01), and base excess (r = 0.68 and 0.71, p < 0.01 and p < 0.001, respectively). The relationship between FEV₁ and Pa_CO2 is shown in Figure 1. There was an inverse relationship between the parameters, with a disproportionate increase in Pa_CO2 in those subjects where FEV₁ was  20% predicted. Maximal inspiratory and expiratory mouth pressures and TLC were not significantly related to gas exchange.

Relationship between Ventilatory Capacity and Sleep Oxygenation

FEV₁ and FVC were significantly related to TST < 90% (r = 0.44 and 0.46, respectively, p < 0.05) while maximal mouth pressures and TLC were not. The relationship between FEV₁ and TST <90% is shown in Figure 2. An FEV₁ of < 40% predicted was 91% sensitive but only 50% specific for a TST < 90% of  2%.

Relationship between Gas Exchange and Sleep Oxygenation

Wakeful Pa_CO2 and base excess were highly correlated with TST < 90% (r = 0.71 and 0.80, respectively, p < 0.001). These relationships are shown in Figure 3. A daytime Pa_CO2 of  45 mm Hg was 91% sensitive and 75% specific for a TST < 90% of  2%. A base excess of  4 mmol/L was 100% specific, but only 55% sensitive for a TST < 90% of  2%. Wakeful Pa_O2 was less closely correlated with TST < 90% (r = 0.52, p < 0.05) than was Pa_CO2.

Using forward stepwise linear regression to examine the relationships of FEV₁, FVC, TLC, maximal mouth pressures, Pa_O2, Pa_CO2, pH, and base excess to sleep hypoventilation, base excess explained 64% of the variability of TST < 90%. The other parameters did not significantly add to the predictive value of base excess.

A linear combination of wakeful Pa_CO2 and TST < 90% was more predictive of base excess (r = 0.93, p < 0.001) than either parameter alone (r = 0.87, p < 0.001 and r = 0.80, p < 0.001, respectively).

AHI

Of all the daytime parameters of ventilatory capacity assessed, the AHI was significantly related to TLC only (r = 0.57, p < 0.01). In particular, it was not related to FEV₁ (r = 0.28), FVC (r = 0.31), Pa_CO2 (r = 0.2), Pa_O2 (r = 0.07), or base excess (r = 0.18).

Effect of NIV

Eight of the subjects assessed were commenced on NIV because of significant sleep-related hypoxemia (TST < 90% of  2%). Three other subjects with a TST < 90% of  2% (2.1, 3.9, and 6.1%) were asymptomatic and declined NIV at the time of polysomnography. Relative to the 11 not commenced on NIV, these 8 subjects had higher wakeful Pa_CO2 (54.0 ± 7.4 versus 43.9 ± 3.5 mm Hg, p < 0.005), higher TST < 90% (32.5 ± 19.6 versus 1.2 ± 2.0%, p < 0.001), lower FEV₁ (19.4 ± 10.1 versus 35.3 ± 18.6% pred, p < 0.05), and lower FVC (18.0 ± 10.3 versus 34.5 ± 16.8% pred, p < 0.05).

Over a mean follow-up period of 0.9 ± 0.4 yr, there was a decrease in Pa_CO2 with NIV (54.0 ± 7.4 to 49.1 ± 4 mm Hg, p < 0.02) despite continued decline in FEV₁ (0.84 ± 0.46 to 0.64 ± 0.39 L, p < 0.05) and FVC (0.91 ± 0.54 to 0.68 ± 0.42 L, p < 0.05). There was no significant change in Pa_O2 (79.8 ± 9.14 to 79.4 ± 7.6 mm Hg, p = 0.94).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Current evidence supports a beneficial role for NIV in DMD once respiratory failure has been established but possibly not in a preventive role before its occurrence. The precise timing of intervention with NIV in this condition needs to be determined. This study was not designed to address this issue directly, but rather to explore the more fundamental question of the relationship between daytime respiratory function and ventilation during sleep and the associated issue of the appropriate timing of polysomnography. This is important, not only for health economic reasons but also to rationalize the use of polysomnography in a group of patients for whom a sleep study is not just an inconvenience but a major organizational event. In addition, denial is common in these patients, who can regard the need for nocturnal ventilation as a preterminal event. The two subjects excluded from this study because of the occurrence of acute respiratory failure requiring NIV had initially refused assessment on the grounds that they felt that they had no respiratory complaints. The ability to identify patients at risk of sleep-related hypoventilation on the basis of spirometry initially may allow more appropriate medical referral by the nurses or physiotherapists with whom these patients have most contact.

We used TST < 90% as our primary indicator of sleep- related hypoventilation, reasoning that hypoventilation of both obstructive and nonobstructive (respiratory pump weakness) origin would be reflected by this measure, given that oxygen therapy was not used in any of our subjects. The concordance between increases in Ptc_CO2 and decreases in Sa_O2 is consistent with the hypoventilatory origin of the desaturations. Previous studies (2, 3) examined relationships to the AHI, that is, to respiratory events defined by criteria more related to upper airway obstruction. This, in part, explains the differences between our findings and these previous studies. Apart from maximum expiratory pressures, Smith and coworkers (2) were unable to identify parameters predictive of sleep hypoventilation in 14 subjects with DMD. However, the range of Pa_CO2 (33.2-45.2 mm Hg) was quite different from that of our subjects (39.5-67.1 mm Hg). Barbe and colleagues (3) identified a significant relationship between AHI and Pa_O2 only, but only six subjects were studied. The age, BMI, and vital capacity of the subjects in that study were similar, but the wakeful Pa_CO2 was lower, than in our subjects. We did not find a significant relationship between AHI and Pa_O2 but did identify an association between the AHI and TLC.

We found that the FEV₁ and FVC were significantly associated with wakeful base excess and Pa_CO2. They were also related to sleep hypoxemia and a FEV₁ < 40% of the predicted value was found to be a sensitive indicator of sleep hypoventilation. However, it is not specific and therefore cannot be used alone to distinguish those with hypoventilation from those without. Base excess and Pa_CO2 had a more direct relationship to sleep hypoventilation. The closer association of base excess than wakeful Pa_CO2 with TST < 90% (r = 0.8 and 0.71, respectively) may be because, besides reflecting acid-base status at the time of measurement, it also reflects buffering of elevated Pa_CO2 during sleep. This suggestion is supported by the fact that base excess was best predicted by a linear combination of TST < 90% and wakeful Pa_CO2, that is, by markers of both wakeful and sleep ventilation. A base excess of > 2 mmol/L was a sensitive indicator of sleep hypoventilation but this threshold is well within the reference range of this measurement, indicating considerable overlap with the normal population. In contrast, a base excess of  4 mmol/L was specific for the presence of sleep hypoventilation. A Pa_CO2 of  45 mm Hg (i.e., presence of wakeful respiratory failure) was a sensitive indicator of sleep-related hypoventilation. These findings suggest that an FEV₁ of > 40% is unlikely to be associated with significant sleep hypoxemia but that arterial blood gas analysis, a less convenient screening test than spirometry, should be performed once FEV₁ falls below 40% of the predicted value. Of such subjects, those with a Pa_CO2  45 mm Hg are at high risk of significant sleep hypoventilation, particularly if the base excess is  4 mmol/L, and should have polysomnography performed. It was notable that these thresholds apply regardless of whether the subjects had obstructive apneas (Figures 1-3). In making this distinction we used the presence of more than four obstructive apneas per hour of sleep to distinguish those with unequivocal obstructive sleep apnea while recognizing that hypopneas exhibited by these and other subjects could have an obstructive element, albeit harder to define. Regardless, the finding that these thresholds were generally applicable suggests that, whether obstructive or nonobstructive in origin, sleep-related hypoventilation will occur at a similar point in the progression of the disease.

Not all subjects with a wakeful Pa_CO2 > 45 mm Hg have significant sleep hypoventilation: two of our subjects with Pa_CO2 measures of 51.3 and 45.5 mm Hg had a TST < 90% of 0.4 and 0%, respectively. Evaluation of ventilation during sleep in this group helps determine the need for NIV and, subsequently, provides a baseline against which to assess its efficacy. Although transcutaneous oximetry during sleep (without oxygen therapy) appears most useful in guiding the need for NIV, other sleep-related data are required. For example, sleep staging is necessary, as sleep hypoventilation may be underestimated without it in patients with poor sleep efficiency or a low proportion of REM sleep, the phase of sleep during which hypoventilation tends to be most profound, as was the case with our subjects. In this regard it is important to note that the proportion of REM sleep observed in those of our subjects with significant hypoventilation was similar to those without. A measure of respiratory effort is also needed to allow obstructive and nonobstructive apneas to be distinguished. This information is important in determining the most appropriate form of ventilatory support, if this proves necessary.

The improvement in wakeful Pa_CO2 in those subjects treated with NIV during sleep implicates sleep hypoventilation in the pathogenesis of respiratory failure in this condition. Furthermore, the finding that wakeful ventilation improved despite continued decline in ventilatory capacity (consistent with further deterioration in respiratory muscle function) suggests impaired ventilatory drive as a possible mechanism for this association, rather than other postulated mechanisms such as respiratory muscle fatigue or decreased lung compliance (6).

In summary, our findings suggest that sleep-related hypoventilation can be predicted by daytime respiratory function in patients with DMD and appears to be implicated in the pathogenesis of respiratory failure. We recommend that arterial blood gases be performed in DMD patients when the FEV₁ is less than 40% of the predicted normal value. Polysomnography should be considered when the Pa_CO2 is  45 mm Hg, particularly if the base excess is  4 mmol/L, as significant sleep hypoventilation is likely to be present, requiring consideration of NIV.