Diaphragm pacing systems for amyotrophic lateral sclerosis / motor neuron disease
Abstract
This is a protocol for a Cochrane Review (Intervention). The objectives are as follows:
The primary objectives were to assess the efficacy and safety of diaphragm pacing for the treatment of people with ALS/MND who develop respiratory insufficiency.
Background
Description of the condition
Motor neuron disease (MND), also termed amyotrophic lateral sclerosis (ALS), is a devastating terminal illness. It is the third most common neurodegenerative disease, with an annual incidence of around 2 cases per 100,000 person years and prevalence of 5 to 8 cases per 100,000 people (O'Toole 2008; Scottish MND Register 1992; Worms 2001). Progressive loss of both corticospinal tract motor neurons (upper motor neurons) and bulbar and spinal motor neurons (lower motor neurons) results in muscle paralysis (Radunovic 2009). The first clinical symptoms of ALS/MND may not appear until up to 80% of motor neurons are lost (Schmiesing 2010), and the average length of survival from symptom onset is only three to five years (Onders 2009a). People with ALS/MND experience progressive weakness affecting the limbs, speech, swallowing and breathing muscles, such as the diaphragm and intercostals. As the diaphragm and intercostal muscles become weak, people with ALS/MND initially experience sleep fragmentation and symptoms of carbon dioxide retention. Common symptoms include early morning headaches, unrefreshing sleep with severe fatigue and sleepiness during the day (McDermott 2008; Turkington 2000). When respiratory muscle weakness becomes severe, those affected can be breathless at rest and are prone to recurrent chest infections. These symptoms have a severe impact on patient quality of life (Bourke 2003; Newsom‐Davis 2001). Severe respiratory muscle weakness is a poor prognostic sign and once the forced vital capacity reaches less than 50% of the predicted value, mortality at nine months ranges from 60% to 100% (Fallat 1979; Stambler 1998).
Description of the intervention
Diaphragm pacing is a technique initially developed for the treatment of respiratory muscle weakness in people with spinal cord injury (DiMarco 2005). Diaphragm pacing has allowed these patients to reduce their time on mechanical ventilation and can even remove the need for it completely (Onders 2005). The NeuRX RA/4 Diaphragm Pacing System is a four channel percutaneous neuromuscular stimulation system. It is currently the only neuromuscular stimulation system available for managing respiratory insufficiency due to ALS/MND. Intramuscular electrodes are implanted into the diaphragm via a laparoscopic surgical procedure. The electrode leads are then tunnelled to an exit site on the chest, where they connect to a small external stimulator box that delivers the stimulus pulses and provides respiratory timing.
How the intervention might work
It is hypothesised that the benefits of diaphragm pacing in people with ALS/MND are the result of restoration of the co‐ordination of respiration, lost as a result of upper motor neuron (UMN) dysfunction. There is also a suggestion that diaphragm pacing conditions the diaphragm muscle. In a healthy diaphragm, slow twitch type I muscle fibres predominate. Disuse and suppression of the diaphragm activity as a result of artificial ventilation has been demonstrated to lead quickly to atrophy and to a predominance of fast type IIb muscle fibres (Levine 2008). Type IIb muscle fibres lead to inefficient, uncoordinated diaphragm contractions. In ALS/MND a similar process is likely to occur, from disuse (secondary to UMN dysfunction) and suppression of diaphragm activity because of non‐invasive ventilation (NIV). Diaphragm pacing may condition the diaphragm with a conversion back to efficient type I muscle fibres (Onders 2009b).
Why it is important to do this review
An important advance in the management of respiratory symptoms in ALS/MND was the demonstration of the beneficial effects of NIV. NIV requires people to wear a mask interface, attached to a small ventilator, over their mouth, nose or face. As they breathe in, the machine gives an extra push of air to support the weak diaphragm, enabling a bigger, deeper breath. Bourke 2006 demonstrated that treatment with NIV led to a significant improvement in survival and quality of life in people with ALS/MND with good bulbar function. These findings were reported in a Cochrane review by Radunovic et al (Radunovic 2009; Radunovic 2013). As experience with NIV has developed, other areas of continuing need have been identified which are not sufficiently addressed by NIV alone, such as mask interface problems, claustrophobia, feeding and communication. Eventually respiratory muscle weakness progresses to a point at which intermittent or overnight NIV is ineffective. Diaphragm pacing is an alternative treatment for managing respiratory insufficiency and may offer benefit over NIV in relation to the issues outlined above. However, there is little information relating to the role of diaphragm pacing in ALS/MND. We plan to systematically search for evidence and assess the effects of diaphragm pacing in ALS/MND to answer the question ‐ does diaphragm pacing offer any additional benefit (over NIV or no treatment or sham treatment) for people with ALS/MND?
Objectives
The primary objectives were to assess the efficacy and safety of diaphragm pacing for the treatment of people with ALS/MND who develop respiratory insufficiency.
Methods
Criteria for considering studies for this review
Types of studies
We will consider all randomised controlled trials (RCTs) and quasi‐RCTs with or without blinding for inclusion. Quasi‐RCTs are those where the rule for allocating participants to treatments is systematic, for example allocation by case record number, date of birth, or by alternation.
Types of participants
We will consider studies involving participants with a clinical diagnosis of ALS/MND based on the El Escorial criteria (Brooks 1994; Brooks 2000), at any disease stage, with any clinical pattern of the condition for inclusion.
Types of interventions
We will consider studies that compare diaphragm pacing to current standard care (NIV), no treatment or sham treatment for inclusion.
Types of outcome measures
Primary outcomes
The primary outcome will be overall survival defined as time from randomisation until death (time‐to‐event data).
Secondary outcomes
Secondary outcomes will include the following.
-
Tracheostomy free survival (time‐to‐event data).
-
Quality of life of the patient: disease specific quality of life measure however defined, for example, Sleep Apnoea Quality of Life Index (SAQLI) at six months; Short Form 36 Health Survey (SF36) at six months; EQ 5D at six months (Brooks 1996; Bourke 2003; Bourke 2006).
-
Quality of life of the carer, however defined, for example, Caregiver Burden Inventory (CBI) at six months (Novak 1989).
-
Adverse events: the proportion of people experiencing adverse events related to diaphragm pacing, in whatever way these are defined in the primary research reports.
Controlled trials which consider diaphragm pacing in patients with respiratory insufficiency due to ALS/MND will be considered for this review. However, no restrictions will be applied to study entry based on the outcome measures used in these studies.
We will include outcomes regardless of how they were collected, for example, using any definitions, assessors, scales and time points. We will describe how outcome measures were defined, if the scale was validated and at what time point outcome data were collected. Only validated outcome measures will be accepted for purposes of meta‐analysis.
Main outcomes for 'Summary of findings' table
We will include a 'Summary of findings' table, with the following seven outcomes.
-
Overall survival (time‐to‐event data).
-
Tracheostomy free survival (time‐to‐event data).
-
Disease‐specific quality of life(e.g. SAQLI at six months).
-
SF36 at six months.
-
EQ 5D at six months.
-
Carer quality of life (e.g. CBI at six months).
-
Adverse events.
In the 'Summary of findings' table(s), we will use the GRADE approach to grade the quality of evidence from the included studies for each of our chosen outcomes. We will base our grades (high, moderate, low or very low) on the presence of: limitations in design and implementation suggesting a high likelihood of bias; indirectness of evidence; unexplained heterogeneity or inconsistency of results; imprecision (wide confidence intervals (CIs)); and publication bias. Where we downgrade the quality of evidence for a given outcome, we will document reasons in footnotes. We will follow the guidance in Chapter 12 of the Cochrane Handbook for Systematic Reviews of Interventions (Schünemann 2011), and prepare the table(s) using GRADEPro software (GRADEpro 2008).
Search methods for identification of studies
Electronic searches
We will search the Cochrane Neuromuscular Disease Group Specialized Register, CENTRAL (The Cochrane Library, current issue), MEDLINE (1966 to present) and EMBASE (1980 to present). We will scan conference abstracts for relevant studies.
Search terms will include: ‘motor neuron disease’ or ‘Amyotrophic Lateral Sclerosis’ combined with ‘diaphragm pacing’ or ‘diaphragmatic pacing’ or ‘phrenic nerve stimulation’ or ‘phrenic nerve pacing’ or ‘electric stimulation therapy’ or ‘ electrodes’ or ‘implanted’ or ‘respiratory paralysis’ or ‘respiratory failure’ or ‘respiratory insufficiency’.
We will also search clinical trials registries such as ClinicalTrials.gov (www.clinicaltrials.gov/), UK Clinical Trials (www.ukctg.nihr.ac.uk), and the WHO International Clinical Trials Registry Platform (ICTRP) portal for ongoing trials (www.who.int/ictrp/en/). We will search the following conference abstracts: MND international symposium; European Respiratory Society (ERS) congress and British Thoracic Society (BTS) conference since January 2005. Additionally we will also search journals of known relevance including ALS Journal, Association of British Neurologists (ABN) Journal, American Academy of Neurologists Journal, European Respiratory Journal, Chest and American Journal of Respiratory Critical Care Medicine, since January 2005. We will check the reference list of all included studies and any reviews for additional studies.
The detailed search strategy for MEDLINE is in Appendix 1.
Searching other resources
We will contact study authors identified via the reference section of any included trials, to identify any additional data whether published or unpublished. We will review other well‐designed observational studies, where the population (ALS/MND), intervention (diaphragm pacing) and outcome (overall survival and adverse events) in question are clearly documented, in the 'Discussion' section of the review. We will identify these (non‐randomised studies) via a search in MEDLINE (January 1966 to the present) and EMBASE (January 1980 to the present). We will do this in order to give a comprehensive descriptive narrative of any non‐randomised data as we believe there are few or no RCTs or quasi‐RCTs in this area.
We will search NHSEED (www.crd.york.ac.uk/crdweb/AboutNHSEED.asp) for any economic evidence, and DARE (www.controlled‐trials.com/) and HTA (www.nets.nihr.ac.uk/search/) for other reviews or assessments for use in the Discusson.
Data collection and analysis
Selection of studies
Two review authors (CM and DH) will independently check the titles and abstracts retrieved by the search for potentially relevant studies. We will then retrieve full‐text articles for all potentially relevant studies. Three authors (CM, DH and CMD) will then independently apply the inclusion criteria to these trials to assess eligibility. The whole group of review authors will resolve any differences of opinion by discussion, in order to reach a consensus. There will be no language limitations. We will document the study selection process in a PRISMA flow chart.
Data extraction and management
Two review authors (CM and DH) will independently extract data onto data collection forms. One author will enter the data into RevMan and the other will check the data entry. We will resolve any discrepancies through discussion or recourse to a third author (CMD). Where there are missing data, these authors will contact the authors of the trials to attempt to obtain further data.
We will extract the following study characteristics.
-
Methods: study design, total duration of study, details of any 'run in' period, number of study centres and location, study setting, withdrawals, and date of study.
-
Participants: N, mean age, age range, gender, severity of condition, diagnostic criteria, baseline characteristics, inclusion criteria, and exclusion criteria.
-
Interventions: intervention, comparison, concomitant medications, and excluded medications.
-
Outcomes: primary and secondary outcomes specified and collected, and time points reported.
-
Notes: funding for trial, and notable conflicts of interest of trial authors.
If the review authors are authors of any primary study, they will not extract data from their own study or studies, or assess their risk of bias. Another author or an editor will extract these data, and check the interpretation against the study report and any available study registration details or protocol.
Assessment of risk of bias in included studies
Two review authors will use the Cochrane 'Risk of bias' tool during data extraction to independently assess the risk of bias in included studies (Higgins 2011). We will consider selection bias (random allocation and allocation concealment), blinding of participants and personnel, blinding of outcome assessors, incomplete outcome data, selective outcome reporting and other sources of bias. Blinding participants and personnel may be difficult to achieve due to the nature of the intervention. As such we will assess both the blinding and incomplete outcome data domains separately for death and all other outcomes (except death). We will grade trials as low risk of bias, high risk of bias or unclear risk of bias (unclear indicating that there is too little information to make a judgement or the risk of bias is uncertain). Where the risk of bias is unclear because of insufficient information, we will attempt to contact the study authors to obtain further information to categorise the study as either high or low risk of bias.
Measures of treatment effect
For time‐to‐event outcomes, we will calculate the hazard ratio (HR) and 95% confidence interval (CI). We will derive estimates of the logHR and its variance from summary statistics reported in primary research, calculated using the spreadsheet published by Tierney and colleagues (Tierney 2007), based on methods described by Parmar and colleagues (Parmar 1998). We will express results for dichotomous outcomes as a risk ratio (RR) with corresponding 95% CI. The number needed to treat for an additional beneficial outcome (NNTB) or the number needed to treat for an additional harmful outcome (NNTH) will be calculated as appropriate. We will calculate the mean difference (MD) and 95% CI for continuous outcomes. Where outcomes are conceptually similar, but measured using different validated instruments, we will calculate the standardised mean difference (SMD) and 95% CI. If there are two of more trials with comparable populations and interventions we will attempt to derive outcome data and use these for meta‐analysis.
Unit of analysis issues
Not applicable.
Dealing with missing data
We believe the outcomes in our review are generally accepted as important and are unlikely to have been withheld from final reports; we consider selective outcome reporting to be unlikely. If selective reporting does occur, we will contact the authors for this information: since pacing for MND has only recently become available, studies and authors are likely to be traceable.
Although overall survival, tracheostomy‐free survival and adverse event data are likely to be complete, there is likely to be substantial missing quality of life data. We will undertake sensitivity analyses to quality of life on the one trial for which we will have access to individual patient data, but will not attempt to impute data for other studies. Missing data will arise for many reasons, and it is problematic to impute pausible values without knowing about the reasons why it is missing. Of particular note is missing quality of life data due to death of participants. Our aim is to assess whether pacing i) prolongs survival and ii) maintains quality of life whilst alive. Therefore, although we could impute a low quality of life score (e.g. zero) for participants who have died, our primary interest is in quality of life among those who are remain alive.
We will address non‐reported outcomes ("trial‐level missing data") with a two‐stage strategy. Firstly, we will search for the study protocol (where possible) to find whether the information was collected. If no protocol is available, or if the information was collected but not reported, we will contact investigators or study sponsors in order to verify key study characteristics and obtain missing numerical outcome data where possible. Missing data may also arise if a trial fails to capture outcome data on all participants ("patient‐level missing data"). This may occur for many reasons, and is most likely to affect quality of life data. We will have access to individual patient data for one trial, allowing us to undertake sensitivity analyses, and may be able to obtain additional information from other study authors. Other than this, we will not attempt to impute missing patient data.
Our preferred analyses are intention‐to‐treat and we will record this in the data extraction. If any studies do not use intention‐to‐treat analysis, we will assess risk of bias accordingly.
Assessment of heterogeneity
We will assess heterogeneity using the I2 statistic (Higgins 2002; Higgins 2003), and Chi2 test provided by the Cochrane statistical software Review Manager (RevMan) (RevMan 2014). We will pool results where studies are sufficiently similar in clinical and methodological terms. With the expectation that the number of studies will be low, meaning that estimates of between‐study variation (tau‐squared) will be less robust, we will use a fixed‐effect model initially. We will use a random‐effects model to obtain pooled estimates if the Chi2 test is statistically significant (P < 0.10), or if the I2 statistic is greater than 50%.
Assessment of reporting biases
If a sufficient number of studies are included in the meta‐analysis, we will use the funnel plot method to investigate potential publication bias. We will incorporate any unpublished results that are obtained. The validity of funnel plots depend on the number and size of available studies as well as the true treatment effect. For this reason we will only use the funnel plot if more than 10 studies are included in a meta‐analysis. We will interpret the results of any funnel plots with caution.
Data synthesis
We will use a broad focus approach, as defined in the Cochrane Handbook for Systematic Reviews of Interventions (Loke 2011), to review any adverse events experienced with diaphragm pacing, in whatever way these have been defined by the included studies. If the review authors agree that there is obvious clinical and methodological heterogeneity amongst eligible studies, such that meta‐analysis seems inappropriate, we will undertake a narrative synthesis.
If there are two of more trials with comparable populations and interventions we will attempt to derive outcome data and use these for meta‐analysis. We will not combine or indirectly compare trials where participants have been randomised differently based on their respiratory insufficiency; where participants’ respiratory eligibility criteria are similar, data will be combined and compared. For time‐to‐event outcomes, we will calculate the pooled HR and 95% CI. We will calculate the pooled RR with 95% CI for dichotomous outcomes. For continuous outcomes, we will calculate the pooled MD and 95% CI or the pooled SMD and 95% CI as appropriate.
Subgroup analysis and investigation of heterogeneity
We will test for heterogeneity across studies. We will perform subgroup analyses to evaluate diaphragm pacing versus no treatment and diaphragm pacing versus sham treatment.
Sensitivity analysis
If we identify heterogeneity, we will perform sensitivity analyses repeating the calculation whilst omitting the trials that have low scores on individual quality items. Sensitivity analyses will be performed where high risk of bias is found due to inadequate random sequence generation or due to missing data (as mentioned in 'dealing with missing data' above). However, we acknowledge there may be insufficient trials to conduct these sensitivity analyses.
