Noninvasive ventilation in ALS

Subjects.

All subjects had definite or probable ALS (El Escorial criteria). Patients with life-threatening co-existing disease, inability to communicate, or aged over 75 years were excluded. Of 25 patients who were eligible, 22 agreed to participate in the trial. Of these, 15 were recruited within 3 months of diagnosis, and 7 were referred because of declining respiratory function or respiratory or sleep-related symptoms. One subject later withdrew owing to difficulty traveling to the center. Ethical approval for the study was obtained, and each subject gave informed consent.

Investigations.

Subjects were assessed periodically following enrollment. A trial of NIV was offered subsequently, if and when the predefined criteria (see below) were satisfied during the first year of the trial. Respiratory muscle function was assessed every 2 months by measurement of vital capacity (VC)12 both sitting and supine, maximum static inspiratory (P_I max) and expiratory (P_E max) pressures,2,13⇓ and sniff nasal inspiratory pressure (SNIP).13 VC was measured using a dry wedge spirometer (Vitalograph, Buckinghamshire, UK). P_I max, P_E max, and SNIP were measured using a handheld meter (Precision Medical, N. Yorkshire, UK). To overcome mouth leaks, an adapted anesthetic full facemask was used in subjects with facial muscle weakness. Measurements are expressed as percentage of predicted.14-16⇓⇓ Daytime arterial blood gases while breathing room air were also measured.

Polysomnography was performed in the hospital every 4 months. Sleep stage, arousals, apneas, and hypopneas were scored manually by standard criteria.17-19⇓⇓ The apnea–hypopnea index (AHI) was calculated as number of events per hour of sleep. Mean and nadir oxygen saturation (S_ao₂) and the proportion of time spent with S_ao₂ < 90% were determined.

QoL was assessed every 2 months using the following generic and specific instruments:table 2) plus one item on change in health. In addition, two summary scores may be calculated, the Mental Component Summary (MCS) and the Physical Component Summary.

1. The Short Form-36 (SF-36; UK version)20 is a widely used generic QoL instrument and is reliable, valid, and responsive in ALS.21 It assesses eight domains of QoL (table 2

2. To assess the consequences of respiratory compromise, we used the Chronic Respiratory Disease Questionnaire (CRQ).22 Dyspnea is assessed on activities identified by the subject, a point of importance in ALS as the pattern of disability varies considerably between subjects. The CRQ has been used successfully in other conditions causing respiratory failure.23

3. To assess the effects of sleep disturbance, we used the Sleep Apnea Quality of Life Index (SAQLI)24 and a symptom scale, the Epworth Sleepiness Scale (ESS).25 The SAQLI was developed to assess QoL in patients with obstructive sleep apnea. Its Symptoms domain uses symptoms identified as important by the subject. The SAQLI also assesses any side effects of treatment. The ESS is a simple self-administered questionnaire assessing daytime sleepiness.

Limb and axial muscle strength was assessed using a muscle score calculated by adding together the MRC clinical scale scores for 22 muscle groups (maximum = 110).26 Functional status was assessed using the disease-specific ALS Functional Rating Scale (ALSFRS).27 Bulbar function was assessed on a 0- to 6-point scale.

The criteria for offering a trial of NIV were determined a priori as any of the following:

1. Orthopnea, defined as breathlessness when supine, attributable to respiratory muscle weakness after exclusion of other potential causes.

2. Daytime sleepiness (ESS > 10) or unrefreshing sleep in the presence of sleep-disordered breathing19 or respiratory muscle weakness (P_I max and SNIP <80% predicted).

3. Daytime P_aco₂ > 45 mm Hg.

4. Nocturnal desaturation (defined as S_ao₂ < 90% for ≥5% of the night).

5. AHI > 10 events/h of sleep.

NIV was established in the hospital using a ResMed VPAP STII pressure-support ventilator in spontaneous/timed mode (ResMed Ltd., Abingdon, UK). Inspiratory (IPAP) and expiratory (EPAP) positive airway pressures were adjusted according to nocturnal oximetry, daytime arterial blood gases, and compliance (mean IPAP = 16 cm H₂O, EPAP = 4 cm H₂O). Rise time, IPAP_max (maximum duration of IPAP), and IPAP_min (minimum duration of IPAP) were adjusted to achieve optimal comfort and efficiency of ventilation. Interfaces used included nasal, oronasal, and total facemasks and mouthpieces (ResMed; Respironics, Bognor Regis, UK; and Puritan Bennett, Bicester, UK). Most subjects with bulbar and facial muscle weakness required an oronasal or total facemask. Ventilator asynchrony, usually due to persisting air leaks, was corrected by limiting inspiratory time (IPAP_max). All subjects were taught assisted cough techniques by an experienced respiratory physiotherapist, tailored to their bulbar and expiratory muscle function. After initiation of NIV, QoL, lung function, and compliance with treatment were assessed at 1 and 3 months and then at two monthly intervals thereafter in the outpatient clinic or the patient’s home. NIV compliance was defined as the average number of hours per 24 hours that the subject was breathing through the ventilator (i.e., “mask on” time) and was recorded automatically by the ventilator. All subjects were followed to death or for at least 26 months following initiation of NIV.

Analysis.

The responses of each domain of QoL were compared in terms of the effect size (mean difference/SD) of the maximum change compared with the value at initiation of NIV in subjects continuing NIV after the initial trial (efficacy analysis). Effect size is the mean number of standard deviations a measure changes by and is classified as follows: small = 0.2 to 0.49, moderate = 0.5 to 0.79, large ≥ 0.8.28 The significance of changes in QoL compared with pretreatment was assessed by paired t-test. The period over which QoL was maintained at or above the pretreatment level was also calculated

The criteria for initiating NIV were compared by intention-to-treat analysis using 1) effect sizes for the SF-36 MCS and Symptoms domain of the SAQLI (the indices showing the greatest improvement in the efficacy analysis) at 1 month to minimize any bias due to a survivor effect; and 2) NIV compliance over the total treatment period (in subjects who were intolerant of, or declined, NIV, compliance during the trial period was calculated).

The outcome for subjects with moderate or severe bulbar impairment compared with those with normal function or mild weakness was similarly assessed by the changes in the SF-36 MCS and SAQLI Symptoms domain at 1 month and NIV compliance. To eliminate the influence of different criteria for initiating NIV, this comparison was restricted to subjects established on NIV for orthopnea (the largest group).

Relations between 1) survival and 2) duration of QoL benefit and each of compliance, age, gender, P_I max, P_aco₂, bulbar score, and limb and axial muscle score were evaluated by univariate and multivariate analysis. Relations between compliance and subject characteristics at initiation of NIV were also assessed. To eliminate the effects of differences in the timing of initiation of NIV, this analysis was also limited to the orthopnea group.

Responsiveness to NIV of QoL domains and duration of benefit.

Large or very large improvements were seen in the ESS, SF-36 (Mental Health and Emotional Limitation domains and MCS), CRQ (Fatigue and Mastery domains), and SAQLI (Social Isolation and Symptom domains, summary score). The CRQ Dyspnea and Emotional Function domains showed moderate improvements. The SF-36 Energy Vitality domain showed a trend for a moderate improvement overall and a large improvement in the nine subjects with orthopnea (effect size = 1.13, p = 0.035). For most domains of QoL, the peak improvement in QoL was seen 3 to 5 months after starting treatment (figure). Responsive domains of QoL were maintained at or above baseline for most of the period of follow-up and survival (median duration for individual domains 252 to 458 days). In contrast, the ALSFRS and domains of QoL assessing physical function declined in all subjects, reflecting disease progression. In general, domains with the largest effect sizes showed more sustained improvements. The largest improvements were in the SF-36 MCS and SAQLI Symptoms domain, and these two indices were therefore chosen to compare the different criteria for initiating NIV (table 2).

• Download figure
• Open in new tab
• Download powerpoint

Figure. Serial data for the Short Form-36 Mental Component Summary (SF-36 MCS), Sleep Apnea Quality of Life Index (SAQLI) Symptoms domain, and ALS Functional Rating Scale (ALSFRS) immediately before starting noninvasive ventilation (NIV) and at 1, 3, 5, and 7 months on NIV. The slight improvement in the ALSFRS score at 7 months is due to a survivor effect.

Survival following initiation of NIV.

Median (range) survival of the 10 subjects who elected to continue NIV was 512 (191 to 793) days. To date, 9 of these 10 subjects have died: Five died during an acute respiratory infection and four because of disease progression. Of these, three elected to discontinue NIV shortly before death.

Among subjects with orthopnea, median survival of the nine subjects who complied with NIV was longer than that of the two who declined or were intolerant of NIV (488 vs 28 days; log rank, p = 0.0004).

Influence of NIV on lung function.

The median rate of decline in VC following initiation of NIV was slower than before treatment (pre 2.52 vs post 1.09% predicted/month; sign test, p = 0.039). There was no evidence of an early detrimental effect of NIV on VC; mean VC immediately prior to initiation of NIV was 48.3% predicted, and after 1 month of treatment, it was 47.2% predicted.

Criteria for initiating NIV.

For the 17 subjects who met the criteria for NIV, the effect sizes of changes in SF-36 MCS and SAQLI Symptoms domain at 1 month and mean compliance with NIV over the treatment period are shown in relation to each indication in table 3. At the time they qualified for a trial, six satisfied only one criterion, five satisfied two criteria, and six satisfied more than two criteria. Results for five orthopneic patients who had a normal daytime P_aco₂ are shown separately. It was noteworthy that 9 of the 11 subjects with orthopnea continued using NIV after the initial trial, with a large improvement in QoL and excellent compliance.

Of 13 subjects with unrefreshing sleep or daytime sleepiness (ESS > 10), 12 accepted a trial of NIV and 9 continued NIV after the initial trial. Compliance was good, but benefit was less than in those with orthopnea. Two of these subjects had a trial for sleep symptoms only; despite significant respiratory muscle and bulbar weakness, both discontinued treatment because of lack of benefit.

All subjects with nocturnal desaturation also had daytime hypercapnia, and results for these criteria are therefore combined. All also had orthopnea, but, of note, the blood gas criteria failed to identify four other normocapnic subjects with orthopnea who continued NIV with benefit.

AHI was unhelpful as a guide to treatment; four of the seven subjects with an AHI of >10 had a trial for this criterion alone, but only one continued using NIV, with poor compliance and no improvement in QoL.

Compared with subjects with normal or only mildly impaired bulbar function, those with moderate or severe bulbar involvement showed lower mean compliance (moderate or severe impairment = 4.9 h/day, normal or mild impairment = 9.1 h/day) and QoL benefit (effect size SAQLI Symptoms domain: moderate or severe impairment = 1.85, normal or mild impairment = 3.28).

Predictors of survival, QoL benefit, and NIV compliance.

Duration of survival correlated with NIV compliance (r = 0.70, p = 0.016) only. In univariate analysis, duration of QoL benefit (SF-36 MCS) correlated with compliance (r = 0.86, p < 0.001) and age(r = −0.61, p = 0.048). However, in multivariate analysis, compliance was the only independent predictor of QoL benefit.

In univariate analysis, NIV compliance correlated with age(r = −0.62, p = 0.042) and upper limb muscle score (r = 0.67, p < 0.05) and showed a trend toward correlation with ALSFRS bulbar score (r = 0.58, p = 0.06) and P_Imax (r = −0.56, p = 0.07). In multivariate analysis, the only independent predictors of compliance were age and upper limb muscle score (adjusted r² = 0.76).