Imai et al. examined the association between the dosing regimen of oral prednisolone (PSL) and the achievement of minimal manifestation status or better on PSL <=5 mg/day lasting >6 months in patients with generalized myasthenia gravis (1). The authors classified 590 patients into high-dose, intermediate-dose and low-dose (n=166) groups, and logistic regression analysis was applied to know the prognosis of patients in low-dose group, by splitting observational period into 1 to 3 years of treatment. The authors concluded that a low-dose PSL regimen with early combination of other treatment options was significantly associated with good prognosis. I have two concerns about their study.

First, the dosing regimen of oral PSL should be considered with caution. Namely, the authors set the maximum dose of oral PSL in each group, and standard treatment schedule was selected after each patient was allocated. Mean daily dose of PSL does not become highest in high-dose group in the study, which happens in the study protocol. In addition, there is a possibility of higher frequency in patients with combination of other treatment options, when patients were registered into low-dose group. As the age of onset was higher and disease duration was shorter in patients with low-dose group, randomized allocation should be strictly conducted in further study.

Second, the number of events was not enough after 1 year observation, and higher odds ratios with wide ranges of confidence interval were observed. As the follow-up period become longer, odds ratios become small, although significance disappeared in comparison between low-dose and high-dose groups after 3 year observation. I suppose that confounders for the prognosis were not completely adjusted, and there is a limitation of study design in multicentre cross-sectional study.

Reference

1. Imai T, Utsugisawa K, Murai H, et al. Oral corticosteroid dosing regimen and long-term prognosis in generalised myasthenia gravis: a multicentre cross-sectional study in Japan. J Neurol Neurosurg Psychiatry 2017 Nov 24. doi: 10.1136/jnnp-2017-316625

Dear Editor,
We read with great interest the article of Boentert et al.1 recently published on this journal. In their paper, the authors describe polysomnographic findings in a large series of non-ventilated patients with amyotrophic lateral sclerosis (ALS). One of the points the authors underscore is that in their patients most respiratory events during sleep were of the obstructive, and not of the central type. This finding is in agreement with what described by Kimura,2 but not by several other authors who found that most respiratory events were of the central type.3 Furthermore, the authors of this paper did not find that obstructive apneas were preferentially associated with bulbar ALS, similarly to David et al.4 but unlike what described by Santos et al.5 However, the tables in the article show that most of the respiratory events were hypopneas, whose type was not specified.
Criteria for scoring sleep disordered breathing events have changed over time, especially as regards hypopneas. Only recently a general rule for type of hypopnea, central or obstructive has been proposed. Obstructive hypopneas are those where at least one of three criteria is met during the event: presence of snoring, flow limitation demonstrated by a flattening of the nasal pressure derived flow signal, opposing thoracic and abdominal movements. Absence of all these criteria characterizes central hypopneas.6 However, standard criteria for apneas and hypopneas fit well to patients with sleep apnea syndromes, but may be difficult to use in patients with other disorders, and in particular in those with neuromuscular diseases.
The separate occurrence of central and obstructive hypopneas was reported in some ALS studies, although standard criteria were not applied. However, a sharp separation of obstructive and central hypopneas in patients with ALS may be difficult, either using standard criteria or not, and often factitious. A clear flattening of the flow signal may not be clearly identified with a weak flow signal. Paradoxical thoraco-abdominal movements may be the effect not only of upper airway narrowing but also of respiratory muscle weakness. Finally, perhaps most importantly, presence of obstruction during a hypopnea does not exclude a simultaneous reduction of respiratory drive, thus a concomitant central genesis of the event.
Even when dealing with apneas, the precise nature of an event in neuromuscular patients is often dubious based on its polygraphic appearance. In the International Classification of Sleep Disorders, central apneas are indicated as the result of a failure of ventilatory control centers to initiate ventilatory effort.7 However, in common practice central apneas are identified by the absence of respiratory movements on polygraphic recordings, although it may not be always easy to precisely identify the pathogenesis of an event based on its polygraphic expression. In neuromuscular patients, many apneas are associated with very small movements. That may lead different scorers to classify them as either central or obstructive. Some authors suggested to introduce the new category of “pseudocentral apneas” to indicate events occurring when the contraction of respiratory muscles is too weak to maintain breathing, which may be more often seen in phasic REM sleep.8 During these events, deflections in the traces of the respiratory movements are very small. Despite their polygraphic aspect resembles apneas more often of the central type, their pathogenesis depends more on peripheral that on central causes. This category of events is not taken in consideration by most authors studying neuromuscular patients. Probably, a non-uniform way to score respiratory events in ALS is one reason for some discrepancies among studies.
We are aware that a correct identification of central and obstructive events is important, because they require different ventilatory approaches that may have a strong influence on sleep quality, therapy acceptance and, in the long-term, disease evolution and survival. Use of positive expiratory pressure, which is mandatory when obstructive events are present, is not only useless, but also questionable in case of absence of upper airway obstruction, as it enhances patient-ventilator asynchronies and sympathetic nervous system tone during sleep.9
Possibly, criteria for scoring sleep disordered breathing events in non-ventilated patients with neuromuscular diseases should be better standardized. They could be useful to have a better agreement among physicians who evaluate polysomnographic recordings, and to better guide in the choice of an appropriate therapeutic strategy.

REFERENCES

1 Boentert M, Glatz C, Helmle C, et al. Prevalence of sleep apnoea and capnographic detection of nocturnal hypoventilation in amyotrophic lateral sclerosis. J Neurol Neurosurg Psychiatry 2017 doi: 10.1136/jnnp-2017-316515.
2 Kimura K, Tachibana N, Kimura J, et al. Sleep-disordered breathing at an early stage of amyotrophic lateral sclerosis. J Neurol Sci 1999;164:37-43.
3 Aboussouan LS, Mireles-Cabodevila E. Sleep-Disordered Breathing in Neuromuscular Disease: Diagnostic and Therapeutic Challenges. Chest 2017;152:880-892.
4 David WS, Bundlie SR, Mahdavi Z. Polysomnographic studies in amyotrophic lateral sclerosis. J Neurol Sci 1997;152:29-35
5 Santos C, Braghiroli A, Mazzini L, et al. Sleep-related breathing disorders in amyotrophic lateral sclerosis. Monaldi Arch Chest Dis 2003;59:160-5.
6 Berry RB, Budhiraja R, Gottlieb DJ, et al. American Academy of Sleep Medicine. Rules for scoring respiratory events in sleep: update of the 2007 AASM Manual for the Scoring of Sleep and Associated Events. Deliberations of the Sleep Apnea Definitions Task Force of the American Academy of Sleep Medicine. J Clin Sleep Med 2012;8:597-619
7 American Academy of Sleep Medicine . International Classification of Sleep Disorders . 3rd ed. Darien, IL : American Academy of Sleep Medicine; 2014.
8 Gould GA, Gugger M, Molloy J, et al. Breathing pattern and eye movement density during REM sleep in humans. Am Rev Respir Dis 1988;138:874-7
9 Crescimanno G, Greco F, Arrisicato S, et al. Effects of positive end expiratory pressure administration during non-invasive ventilation inpatients affected by amyotrophic lateral sclerosis: A randomized crossover study. Respirology 2016;21:1307-13

In our previous meta-analysis of cerebellar atrophy in seven major neurodegenerative conditions (Alzheimer’s disease (AD), amyotrophic lateral sclerosis (ALS), frontotemporal dementia (FTD), Huntington’s disease (HD), Parkinson’s disease (PD), multiple system atrophy (MSA), and progressive supranuclear palsy (PSP)) we investigated studies that reported grey matter (GM) loss in the cerebellum [1]. Consistent regions of atrophy were found in AD, ALS, FTD, MSA, and PSP but not HD or PD. In their comment on our meta-analysis, Sheng and Pan have argued that our method of selectively investigating studies that found cerebellar atrophy, rather than adopting a whole-brain approach, is “not optimal in a coordinate-based meta-analysis” [2]. They further cite previous whole-brain meta-analyses that did not identify clusters of cerebellar grey matter loss in patients [3-6].
Here, we argue that our approach was justified given our aim, which was to focus on cases where cerebellar atrophy was found in the respective disease groups in order to determine 1) if such atrophy followed a consistent, robust pattern, and 2) if atrophy patterns were disease-specific or generic, where possible relating them to symptomatology.
There are several reasons why we chose to focus on the cerebellum rather than adopting a whole-brain approach. First, the cerebellum is hardly a “region of interest” in the classical sense given its marked heterogeneity in terms of function and connectivity as well as its large cell count [7,8]. Using an ROI approach on the cerebellum is therefore not vastly different from conducting an analysis restricted to the neocortex, a practice that is abundant and well accepted. The vast majority of coordinates used in our analysis stemmed from whole-brain analyses and omitting the few studies with an ROI approach for the cerebellum is highly unlikely to affect the main clusters we identified.
Second, there is an abundance of coordinate-based meta-analyses using ROI approaches (see http://www.brainmap.org/pubs/), including one of the most influential papers into the functions of the cerebellum [8].
Third, the systematic literature search of papers investigating the diseases in question has revealed that many analyses were inherently biased against the cerebellum. As shown in our PRISMA flowchart, there was a tendency towards excluding the cerebellum (n=64) or not reporting or discussing cerebellar findings in text even when figures indicated the cerebellum was affected (n=13 for which we could not obtain data even after contacting authors). Twenty-seven studies further indicated potential cerebellar effects but did not report coordinates. These numbers made it clear that this bias in investigating the diseases of interest and in reporting results would inadvertently be translated into a bias in the meta-analytic results when choosing a whole-brain approach.
Given this abundance of studies excluding the cerebellum and underreporting cerebellar findings, it is not surprising that previous meta-analyses did not find robust clusters of cerebellar atrophy. Furthermore, there are additional factors that can explain the apparent lack of GM loss in the cerebellum in the AD, ALS, and FTD meta-analyses.
First, the majority of studies in AD and ALS investigated early stages, during which cerebellar atrophy may not yet be apparent. This tendency to investigate early structural changes may contribute to the belief that the cerebellum is typically spared. Given that clinical assessment using magnetic resonance imaging is carried out early in disease progression and that follow-up scans in clinical practice are rare, later GM loss often remains unexplored. However, neuropathological investigations have found the cerebellum to be affected in AD, ALS, and PSP [9-11]. Second, in our systematic literature search, we contacted authors to obtain additional coordinate data, which was not done in most of the other meta-analyses. The following should emphasize these arguments:
• AD meta-analysis [3]: did not include seven studies present in our analysis; of the 46 AD studies listed in Table 1, n=23 included exclusively early-stage/mild AD, and n=9 studies mild to moderate AD. One study with preclinical AD cases and another with a sample of MSA with dementia would not have met our inclusion criteria [12,13].
• FTD meta-analysis [4]: did not include nine studies from our analysis. 3/9 studies in the whole-brain analysis showed cerebellar atrophy; of the remaining six, n=3 included only patients with mild FTD severity, while n=1 included mild to moderate.
• ALS meta-analysis [5]: did not include two of the studies in our paper. 5/20 studies reported disease-related cerebellar grey matter (GM) or white matter integrity loss; of the other 15, n=10 focussed on patients with mild symptom severity. Interestingly, Minnerop et al. showed a correlation of cerebellar GM loss and disease duration, suggesting that the cerebellum is affected later in the course of ALS [14].
• We would like to point out that the PSP meta-analysis [6], which was cited by Sheng and Pan as not having detected cerebellar GM loss, did find a cluster of atrophy in right declive (see Table 2). This meta-analysis also did not include two of the studies incorporated in our investigation. Cerebellar GM loss was found in 5/12 studies in this meta-analysis, while n=4 of the remaining seven found white matter loss in cerebellar peduncles in the absence of GM differences. Of the three studies with no cerebellar findings, n=2 were in mild PSP.

Finally, we would like to emphasise that we are not claiming that cerebellar atrophy is necessarily present in the majority of patients in all of the investigated diseases. Given that we only selected studies with findings in the cerebellum, we did not seek to determine the likelihood of the cerebellum being affected in a given disease. For our purpose, a focus on studies with cerebellar findings is justified. We conclude that if cerebellar atrophy is present, it follows a disease-specific pattern that tends to correlate with symptomatology in many of the included studies. Our findings demonstrate the importance of future research into the role of the cerebellum in neurodegeneration given the two major issues in the literature which are the tendency to disregard the cerebellum and the vast heterogeneity of patient groups regarding disease stages and subtypes.

References
1. Gellersen HM, Guo CC, O’Callaghan C, et al. Cerebellar atrophy in neurodegeneration—a meta-analysis. J Neurol, Neurosurg Psychiatry 2017, in press.
2. Sheng L-Q, Pan P-L. Does Cerebellar Atrophy in Neurodegeneration? J Neurol, Neurosurg Psychiatry 2017. http://jnnp.bmj.com/content/early/2017/05/27/jnnp-2017-315607.responses
3. Chapleau M, Aldebert J, Montembeault M, Brambati SM. Atrophy in Alzheimer's Disease and Semantic Dementia: An ALE Meta-Analysis of Voxel-Based Morphometry Studies. J Alzheimers Dis 2016;54(3):941-55.
4. Schroeter ML, Laird AR, Chwiesko C, et al. Conceptualizing neuropsychiatric diseases with multimodal data-driven meta-analyses - the case of behavioral variant frontotemporal dementia. Cortex 2014;57:22-37.
5. Sheng L, Ma H, Zhong J, et al. Motor and extra-motor gray matter atrophy in amyotrophic lateral sclerosis: quantitative meta-analyses of voxel-based morphometry studies. Neurobiol Aging 2015;36(12):3288-99.
6. Yu F, Barron DS, Tantiwongkosi B, Fox P. Patterns of gray matter atrophy in atypical parkinsonism syndromes: a VBM meta-analysis. Brain Behav 2015;5(6):e00329.
7. Balsters JH, Laird AR, Fox PT, Eickhoff SB. Bridging the gap between functional and anatomical features of cortico-cerebellar circuits using meta-analytic connectivity modeling. Hum Brain Mapp 2014;35(7):3152-69.
8. Stoodley CJ, Schmahmann JD. Functional topography in the human cerebellum: A meta-analysis of neuroimaging studies. NeuroImage 2009;44(2):489-501.
9. Kanazawa M, Shimohata T, Toyoshima Y, et al. Cerebellar involvement in progressive supranuclear palsy: A clinicopathological study. Mov Disord 2009;24(9):1312-8.
10. Larner AJ. The Cerebellum in Alzheimer's Disease. Dement Geriatr Cogn Disord 1997;8(4):203-9.
11. Prell T, Grosskreutz J. The involvement of the cerebellum in amyotrophic lateral sclerosis. Amyotroph Lateral Scler Frontotemporal Degen 2013;14(7-8):507-15.
12. Hämäläinen A, Pihlajamäki M, Tanila H, et al. Increased fMRI responses during encoding in mild cognitive impairment. Neurobiol Aging 2007;28(12):1889-903.
13. Tondelli M, Wilcock GK, Nichelli P, et al. Structural MRI changes detectable up to ten years before clinical Alzheimer's disease. Neurobiol Aging 2012;33(4):825.e25-36.
14. Minnerop M, Specht K, Ruhlmann J, et al. In vivo voxel-based relaxometry in amyotrophic lateral sclerosis. J Neurol 2009;256(1):28-34.