Respiratory function vs sleep-disordered breathing as predictors of QOL in ALS

Subjects.

All patients with newly diagnosed ALS and those with established disease referred due to declining respiratory function or respiratory/sleep related symptoms attending a regional ALS center were invited to participate. We recruited 23 subjects, of whom 15 underwent their first assessment within 3 months of diagnosis (duration of weakness 3–37 months). Patients with other serious comorbidity, such as cancer or unstable ischemic heart disease, or aged over 75 years were excluded. All patients referred due to declining respiratory function or symptoms agreed to participate; three recently diagnosed patients declined. All subjects had definite or probable ALS by the El Escorial criteria.

Investigations.

Respiratory muscle function was assessed by measurement of vital capacity (VC) sitting and supine, maximum static inspiratory (P_I max) and expiratory (P_E max) pressures, and sniff nasal inspiratory pressure (SNIP).1-3,7-9⇓⇓⇓⇓⇓ VC was measured using a dry wedge spirometer (Vitalograph, Buckingham, England). An adapted anesthetic full-face mask was used for patients with bulbar weakness in order to overcome mouth leaks. Maximum static pressures were measured using a pressure transducer (Validyne, Northridge, CA), flanged mouthpiece, and nose clip with attention to obtaining a good lip seal. Measurements were expressed as % of predicted normal values.7-9⇓⇓ Daytime arterial blood gases breathing room air were measured. A summated muscle score based on the Medical Research Council clinical scale was used to estimate limb and axial muscle strength. This gives a score between 0 and 110 (normal). It has been widely used and shown to be a responsive index of functional status.10

Polysomnography was performed with manual scoring of sleep stage in 30-second epochs, apneas, hypopneas, and arousals by standard criteria.11-13⇓⇓ The apnea hypopnea index (AHI) and arousal index were calculated as number of events per hour of sleep. Oxygen saturation (SaO₂) was monitored continuously, and mean and nadir values and the proportion of time spent with SaO₂ <90% were determined.

QOL was assessed using both generic and specific instruments. Two self-administered generic instruments were used: the Short Form–36 UK version (SF-36)14 and the general well-being schedule (GWB).15 The SF-36 is the most widely used instrument of its kind and has been shown to be reliable, valid, and responsive in a wide range of conditions including ALS.16 It assesses eight domains of QOL (see following), plus one item on change in health. The GWB is a 22-item questionnaire assessing subjective feelings of psychological well-being and distress.

Respiratory problems were assessed using the Chronic Respiratory Disease Questionnaire (CRDQ).17 Although not specifically developed for use in patients with respiratory failure, it has been used successfully in that situation.18 The influence of sleep related symptoms was assessed using the Sleep Apnea Quality of Life Index (SAQLI)19 and the Epworth Sleepiness Scale (ESS).20 The SAQLI was developed to assess QOL in patients with sleep disruption due to obstructive sleep apnea. The ESS is a simple self-administered eight-item questionnaire assessing daytime sleepiness. The disease-specific ALS Functional Rating Scale (ALSFRS) was used to assess fine motor, gross motor, bulbar, and respiratory function.21

Statistical analysis.

The relations of QOL to respiratory muscle function and polysomnographic indices were examined by Spearman rank correlation coefficients. Subsequently a multiple regression analysis was performed with casewise deletion of missing values to identify independent predictors of QOL. Only those variables with evidence of predictive value following univariate analysis were included in the model. The analysis was repeated on the recent diagnosis group alone, as this subgroup was otherwise entirely unselected.

Respiratory muscle function.

One subject with severe bulbar and respiratory involvement was unable to perform the P_I max maneuver. The majority of subjects had evidence of respiratory muscle weakness at enrollment (20/23). All measurements of respiratory muscle strength tended to be lower in patients with moderate or severe bulbar weakness (P_I max: p = 0.087). Seven subjects had daytime hypercapnia (median PaCO₂ = 41.0 mm Hg, range = 34.3–113 mm Hg). Results of the summated muscle score are shown in table 1.

Polysomnography.

One subject with established ALS was referred with severe hypercapnia (PaCO₂ = 113 mm Hg while breathing oxygen at 2 L/min⁻¹ by nasal cannulae) and drowsiness. Baseline polysomnography was not performed and noninvasive ventilation was instituted immediately. Another subject with a long history of orthopnea and symptoms of sleep disruption had predominantly alpha-delta activity on EEG, such that conventional sleep staging was not possible. The median AHI was 7.7 events/hour sleep, and nine patients had an AHI greater than 10. Five subjects spent >5% of the night with SaO₂ <90%. All had severe muscle weakness with P_I max <30% predicted. The disruption to sleep architecture, as illustrated by the arousal index and proportions of stage one and REM sleep, was more striking than respiratory events, and was seen even without a raised AHI. Compared to normal values,22 total sleep time and REM sleep were both reduced, with complete suppression of REM in four subjects with severe muscle weakness. The duration (but not proportion) of slow wave sleep was also reduced. Stage one sleep and EEG arousal index were increased.

Relationship between QOL and respiratory muscle function.

SF-36.

There were highly significant, moderate to strong correlations between four of the eight domains of QOL within the SF-36 (general health perception, energy vitality, physical function, and social function) and all measurements of respiratory muscle strength (R = 0.42–0.73) (table 2). We encountered major “floor” and “ceiling” effects in the domains of role limitation due to physical and emotional problems. No significant correlations were seen between respiratory muscle function and the role limitation or pain domains. The mental health domain correlated with P_I max only. The relationship between the general health perception domain and P_I max is illustrated (figure).

View this table:

• View inline
• View popup

Table 2.

Univariate analysis of quality of life (QOL) with respiratory muscle function and muscle score

• Download figure
• Open in new tab
• Download powerpoint

Figure. The relationship between Short Form–36 general health perception and P_I max. Spearman correlation R = 0.73, p < 0.0002.

Relations between QOL and (limb and axial) muscle score.

The strongest correlation was between muscle score and the physical function domain of the SF-36 (R = 0.84). Muscle score also correlated with the ALSFRS. The only other significant correlations were with the social function and general health perception domains of the SF-36. In particular, there were no significant relations between muscle score and instruments specific to sleep or respiratory related problems.

Relations between QOL and polysomnography.

The relations between QOL and polysomnographic indices were weaker and less consistent than those between QOL and respiratory muscle function. Mean SaO₂ correlated with the ESS (R = −0.50), SF-36 general health perception (R = 0.50), and the CRDQ mastery and dyspnea domains (R = 0.56). Total sleep time correlated with SF-36 physical function (R = 0.47), CRDQ mastery (R = 0.44), SAQLI summary score (R = 0.47), and the ALSFRS (R = 0.59). Percentage stage one sleep correlated with SF-36 pain (R = 0.50). Percentage REM sleep correlated with the CRDQ mastery and dyspnea domains (R = 0.48) and the ALSFRS (R = 0.46). No other significant correlations were observed with any other polysomnographic variables, including AHI.

Independent predictors of QOL.

We performed standard multiple regression analysis with casewise deletion of missing values to identify independent predictors of QOL. The analysis was performed separately for the ALSFRS and each of the three domains of the SF-36 and CRDQ that correlated most strongly with measurements of respiratory muscle strength and polysomnographic indices. For each QOL measure only those indices with predictive value on univariate analysis were included in the multiple regression model (two to four independent variables). In univariate analysis there were significant relations between all measurements of respiratory muscle strength and QOL, but this was generally strongest for P_I max and this was therefore used in the multiple regression model. The other variables included following univariate analysis were as follows: daytime PaCO₂, mean nocturnal SaO₂, total sleep time, % REM, and muscle score.

For the three domains of the SF-36 and CRDQ, multiple regression analysis showed that P_I max was the only independent predictor of QOL. Both P_I max and muscle score were independent predictors of the ALSFRS. However, when analyzed in a forward stepwise model, most of the predictive value of the model was accounted for by P_I max (table 3). No polysomnographic index was an independent predictor of QOL or of functional status. The observed relations between QOL and all polysomnographic indices examined were entirely explained by coexistent respiratory muscle weakness.

Respiratory muscle function.

The high prevalence of respiratory muscle weakness is consistent with current literature in ALS.1 Maximum static pressures and SNIP were more sensitive indicators of respiratory muscle weakness than VC, as would be expected from the curvilinear pressure volume relationship of the respiratory system. Patients with moderate/severe bulbar weakness tended to have more severe respiratory muscle weakness (p = 0.087) and less severe limb and axial muscle weakness. The difference is unlikely to be due artifactually to mouth leakage as this was minimized in appropriate patients by use of a full-face mask during testing.

Polysomnography.

Sleep-disordered breathing has been reported in between 17% and 76% of patients with ALS.4-6,23⇓⇓⇓ However, in most previous studies subjects were selected on the basis of symptoms or bulbar involvement. Sleep disordered breathing has been reported in subjects with ALS and bulbar impairment at an early clinical stage and without symptoms of respiratory compromise or sleep disturbance.23 Our study is the first to report polysomnography in unselected recently diagnosed patients with ALS. Of 15 such individuals, five had an AHI >10, although in two subjects this might have been unrelated as they had no evidence of bulbar or respiratory involvement. Of the remaining 10 recently diagnosed subjects, eight had disturbance of sleep architecture without a raised AHI. Our results provide convincing evidence that both sleep disruption and sleep disordered breathing are common at the time of diagnosis of ALS. For the whole group, the disruption of sleep architecture and the proportion with an AHI >10 were greatest in patients with the most severe respiratory muscle weakness. The most striking observation was the complete suppression of REM sleep in four subjects, all of whom had a P_I max <30% predicted. Suppression of REM sleep in ALS subjects with severe diaphragmatic weakness has recently been reported, and as the intercostal and accessory muscles are not normally active during REM sleep, this may represent a compensatory mechanism.24

Assessment of QOL.

The SF-36 has been validated, and currently is the instrument of choice for assessing QOL in ALS.25 No QOL instrument that specifically assesses the impact of respiratory or sleep related problems has been validated in ALS. We chose the CRDQ to assess respiratory problems because it has been used successfully in subjects with respiratory failure, and the dyspnea domain is assessed on activities selected by the subject, which is important as the pattern and degree of disability is very variable in ALS. The questions in the other domains (fatigue, emotional function, and mastery) are also relevant to aspects of ALS. The SAQLI was chosen as the symptoms domain is assessed on sleep related symptoms that the subject has identified as being most important. A few questions in the other domains are not relevant to this population, and such items can be omitted (these domains contain 11–13 items and are scored by calculating the average for the questions answered). There were strong correlations between related domains in the SF-36, GWB, CRDQ, and SAQLI, supporting our use of these instruments.

Relations between overall and respiratory muscle strength, polysomnography, and QOL.

The degree of respiratory muscle weakness was related not only to symptoms of orthopnea and dyspnea, but also to unrefreshing sleep and daytime somnolence. The strength and consistency of the relations between respiratory muscle function and both QOL and functional status were striking. There was a moderately strong correlation between respiratory muscle strength and dyspnea as assessed by the CRDQ. Furthermore, P_I max was the strongest independent predictor of QOL in a multiple regression model including polysomnographic indices and limb and axial muscle score. Although the AHI was elevated, it did not correlate with any domains of QOL, and no polysomnographic index was an independent predictor of QOL. Significant disruption of sleep architecture was seen in subjects with respiratory muscle weakness even in the presence of a normal AHI. Our results therefore suggest that symptomatic sleep disruption is attributable predominantly to respiratory muscle weakness, and apneas and hypopneas are of little importance in determining QOL. They also cast doubt on the need for routine sleep studies in such patients, provided that adequate daytime assessment is performed.

Intuitively one would expect overall muscle score to have been an independent predictor of the SF-36 physical function domain, as observed with the ALSFRS. However, we found significant floor effects in the physical function domain of the SF-36 (similar to the role limitation domains), which were not seen with the ALSFRS. Our results therefore suggest that the disease-specific ALSFRS is simply a more sensitive instrument in this respect.

Whereas the 15 recently diagnosed patients were unselected (other than by age <75 years and absence of serious comorbidity), the other patients were referred on the basis of symptoms or declining respiratory muscle function, which may have biased the results. We therefore repeated the multiple regression analysis including only the recently diagnosed subjects. P_I max remained the strongest predictor of QOL, and again no polysomnographic index was found to be an independent predictor of QOL.

In disabling conditions, respiratory and sleep related symptoms may easily be overlooked clinically, and under-reported by patients.2,26⇓ The high prevalence of respiratory muscle weakness and sleep disordered breathing in ALS, even at diagnosis, emphasizes the importance of careful clinical assessment combined with simple monitoring of respiratory function with maximum respiratory pressures and VC. The strong relationship between respiratory muscle strength and QOL raises hope that supportive treatment with noninvasive ventilation (NIV) may substantially improve QOL. NIV has been shown to improve survival in ALS,27,28⇓ but the impact on QOL is unclear. In a distressing, progressive, and incurable condition such as ALS, QOL is arguably a more important outcome measure than survival. There is an urgent need for further trials to establish the optimal criteria (if any) for initiating NIV in ALS, and the impact on QOL.