Pulmonary function tests (maximum inspiratory pressure, maximumexpiratory pressure, vital capacity, forced vital capacity) predict ventilator

use in late-onset Pompe diseaseErin M. Johnson a,*, Mark Roberts b, Tahseen Mozaffar c, Peter Young d, Adrian Quartel a,

Kenneth I. Berger e

a BioMarin Pharmaceutical Inc., Novato, CA, USAb Department of Neurology, Salford Royal NHS Foundation Trust, Salford, UK

c Department of Neurology, University of California, Irvine, CA, USAd Department of Sleep Medicine and Neuromuscular Disorders, University Hospital Münster, Münster, Germany

e Departments of Medicine, Physiology and Neuroscience, New York University School of Medicine, New York, NY, USA

Received 26 June 2015; received in revised form 23 November 2015; accepted 25 November 2015

Abstract

In patients with Late-Onset Pompe Disease (LOPD), progressive respiratory muscle involvement leads to reduced pulmonary function, withrespiratory failure the most common cause of mortality. Early disease manifestations include sleep-disordered breathing, which can be treated withnon-invasive ventilation; however, progressive diurnal deficits can require invasive ventilation. To determine if pulmonary function tests (PFTs)predict the thresholds for ventilation and wheelchair use, a systematic literature review identified cross-sectional clinical patient data (N = 174) thatwas classified into ventilation and wheelchair cohorts. PFTs included maximum inspiratory pressure (MIP), maximum expiratory pressure (MEP),forced vital capacity (FVC), and vital capacity (VC), with vital capacities measured in the upright (-U) and supine (-S) positions. Receiveroperating characteristic (ROC) curves were used to calculate cut-points (CP) and area under the curve (AUC). For all ventilation and mobilitythresholds tested, ROC analyses demonstrated AUC values from 86–89% for MIP, 72–96% for MEP, and 74–96% for all vital capacity metrics.Thus, PFTs are useful in predicting the thresholds for nighttime ventilation, daytime ventilation, and wheelchair use, with MIP and VC-U havingboth high AUC values and consistency. The PFT mobility CPs were low (MIP CP = 0.9 kPa, MEP, CP = 2.6 kPa, VC-U CP = 19% predicted),suggesting an endurance component associated with wheelchair use.© 2015 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Keywords: Late-Onset Pompe disease; Inspiratory pressure; Vital capacity; Ventilator; Ventilation; Wheelchair test

1. Introduction

Late-Onset Pompe Disease (glycogen storage disorder type 2,acid maltase deficiency, LOPD) is a rare genetic disordercharacterized by attenuated functionality of the enzyme acid alpha-glucosidase. The resulting accumulation of glycogen in skeletaland respiratory muscles can result in myopathy, with progressivedeterioration in both mobility and pulmonary function. As aheterogeneous disorder, clinical disease can present throughoutlife, from young childhood (>2 years) through adulthood. In over70% of LOPD patients, respiratory failure is the most commoncause of mortality [1,2]. Longitudinal quality of life surveys have

demonstrated reductions in the physical health domains in astep-wise manner with the addition of respiratory support(nighttime and daytime ventilators) and mobility assistancedevices (e.g. wheelchairs) [3–5].

With progressive decline of critical respiratory muscle function(e.g. diaphragm), sleep disordered breathing may arise withassociated symptoms of morning headache, daytime somnolenceand fatigue [6,7]. Studies on sleep in patients with neuromusculardisorders (congenital muscular dystrophy, Duchenne musculardystrophy, spinal muscular atrophy type I-II, limb-girdledystrophy, late-onset Pompe disease/acid maltase deficiency,nemaline myopathy, hereditary motor and sensory neuropathy typeI, centronuclear myopathy, myotonic dystrophy, and non-classifiedmyopathy) have suggested that the onset of both sleep-disorderedbreathing (SDB) and nocturnal hypoventilation (NHV) can bepredicted with daytime pulmonary function measurements (e.g.

* Corresponding author. BioMarin Pharmaceutical Inc. 105 Digital Drive,Novato, CA 94949 USA. Tel.: +1 415 455 7665.

E-mail address: ejohnson@bmrn.com (E.M. Johnson).

http://dx.doi.org/10.1016/j.nmd.2015.11.0090960-8966/© 2015 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Available online at www.sciencedirect.com

Neuromuscular Disorders 26 (2016) 136–145www.elsevier.com/locate/nmd

ScienceDirect

inspiratory vital capacity, IVC; maximum inspiratory pressure,MIP) [8–11]. In neuromuscular disorders, the introduction ofnighttime ventilation has been shown to significantly improve therestfulness of sleep (daytime sleepiness; morning headache) andcorrect daytime hypoxemia and hypercapnia [12,13].

In addition to sleep studies, pulmonary function tests (PFT),specifically MIP and FVC, have been used to establish the needand financial reimbursement for nocturnal mechanical ventilationin patients with neuromuscular disorders [14,15]. While the healthinsurance systems in the United States and the United Kingdomhave defined PFT criterion for reimbursement of nocturnalventilation [15,16], these have not been adopted internationallyand there are no proposed criteria for daytime ventilation. Sincethe implementation of nocturnal, and subsequently, daytimeventilation are important milestones in the progression of LOPD,the present study was designed to evaluate if non-invasivepulmonary function tests (PFTs) can be used to describe thepredictive thresholds for both nocturnal and diurnal ventilatorusage. In addition, as respiratory function can affect endurance andmobility, this study also examined the ability of PFTs to describethe predictive threshold between a mobile state and wheelchairusage.

2. Methods

2.1. Search strategy

Clinical studies reporting both pulmonary function metrics andventilator use of individual LOPD patients were identified using

electronic searches of both EMBASE and MEDLINE databaseon October 7, 2014. The search terms included the followingterms: (“pompe” OR “glycogen storage disease type 2” OR “acidmaltase” OR “glycogenosis”) AND (“respiratory” OR “ventilator”OR “pulmonary”). Using the SCOPUS database, additionalstudies were identified from subsequent articles that had citedthe initial list of clinical studies that had passed the screeningprocess.

To identify only individuals with late-onset disease, clinicalstudies and case reports of patients ≥2 years with both pulmonaryfunction test (PFT) results and a description of ventilator usagewere included. Exclusion criteria also comprised studiesreporting on animal models, infantile Pompe disease (<2 years),other diseases, disease management guidelines, diagnostic tools,genotype-phenotype studies, and articles written in languagesother than English. The selection criteria, shown in Fig. 1,identified 34 studies with patient level data reporting at least 1PFT and a description of ventilator status [6,17–49].

2.2. Data identification and analyses

Individual data were identified from the included clinicalstudies. Averaged values were not used. Pulmonary function dataincluded two measures of respiratory muscle strength, maximuminspiratory pressure (MIP), and maximum expiratory pressure(MEP), as well as four measures of lung volume, vital capacity inthe upright (VC-U) and supine (VC-S) positions, forced vitalcapacity in the upright (FVC-U) and supine (FVC-S) positions.

Fig. 1. Systematic literature review to identify clinical data for pulmonary function tests and ventilator status for patients with late-onset Pompe disease.

137E.M. Johnson et al. /Neuromuscular Disorders 26 (2016) 136–145

The respiratory muscle strength measurements (MIP, MEP) wereanalyzed in terms of pressure (kPa), while the lung capacitymetrics (VC, FVC) were analyzed as percent of predicted values(% predicted). The respiratory muscle strength measurements aretypically reported as absolute values and not normalized for ageor gender. These metrics were the most commonly reported testsin the literature, providing sufficient data for these analyses. Datawere excluded if descriptions of ventilator status for individualpatients were inadequate. Clinical data from patients who weretreated and untreated with enzyme replacement therapy (ERT)were included, with the assumption that treatment may alter thetrajectory of progression but not the relationship between eachPFT parameter and ventilator/ambulation status. To assess thedegree to which treatment (or lack thereof) may influence theseanalyses, a statistical comparison of the treated vs. non-treatedpatients within each cohort was also performed.

Ventilator status was determined by either a writtendescription in the study (no ventilation, nighttime ventilation,or daytime ventilatory assistance) or a description of theventilator time usage per day (with ≤ 12 hours representingnocturnal ventilation and > 12 hours representing daytime use).Preference between these two methods was given to the writtendescription in the text. To incorporate the possible addition ofventilation during longitudinal progression, up to 4 data pointsper individual were included, with a minimum of 3 monthsbetween measurements.

To determine the ability of PFTs to predict the thresholds inventilator status, clinical data were assigned to 1 of 3 cohorts, asdefined with no ventilation (none), nighttime ventilation (night),or daytime ventilation (day). These cohorts were analyzed usingreceiver operating characteristic (ROC) curves, determining thecut point (CP, from the maximum likelihood ratio), the area underthe curve (AUC), as well as the sensitivity and specificity values.Analyses were performed with GraphPad Prism version 6.05(GraphPad Software, La Jolla, CA, USA). Each PFT wasconsidered separately, evaluating the hypothesis that each metriccould independently predict these threshold points. AUC valuesof 100% represent a perfect differentiation between the groups.

In addition, an analysis using only a single point from eachpatient was also performed to determine if there was an inherentbias with the inclusion of longitudinal data. As VC and FVC havebeen used interchangeably in the literature, additional analyseswere performed using combined data sets for VC-U and FVC-U,as well as VC-S and FVC-S. For individual patient data reportingboth a VC and an FVC metric, the larger value was used in thecombined analyses.

To determine the ability of PFTs to differentiate ambulationand wheelchair use, well-characterized clinical data wereassigned to 1 of 2 cohorts (mobile or wheelchair) and alsoanalyzed using ROCs. Wheelchair status was defined by a writtendescription of wheelchair usage, a description stating that a six-minute walk test (6MWT) was not feasible, or a Waltonscore ≥ 7. Mobile status was defined by a written description ofmobility, a value for the 6MWT, or a Walton score < 7. Individualpatient data with inadequately characterized descriptions ofmobility were classified as unknown and not used for thismobility analysis.

3. Results

3.1. Patient characteristics

From the 34 included studies, clinical data describingventilator status and at least one PFT were identified for 174individual patients (N), with characteristics shown in Table 1.With the inclusion of up to 4 data points per individual patientwith at least a 3-month separation between observations, a totalnumber of observations (n) of 267 were identified and usedin these analyses. The ages of the patients at the time ofobservation are also shown in this table, as well as the countriesthat the studies originated in. As these reported data had beende-identified, there was the possibility that individual patientdata may have been represented in multiple reports. Todetermine if there were duplicate records, patient data indifferent studies from the same authors and/or within the samecountry were compared using gender and age. Using thismethod, there were no individual data that overlapped. Toevaluate a possible effect that treatment may have on theseanalyses, the clinical data points were also characterized asuntreated or treated with ERT. Within each PFT (MIP, MEP,VC-U, FVC-U), data were grouped according to treatment andanalyzed with a student t-test to determine if there werestatistically relevant differences that may affect the ROCanalyses. For the observational data, 164 data points wereobtained for untreated and 78 data points were identified asERT treated. The remaining undefined data were not used inthis analysis. In comparing the treated vs. untreated data, thep-values from the student t-test for the three ventilation cohorts(none; night; day) were calculated for MIP (0.55; 0.96; 0.23),

Table 1Patient characteristics.

Individual patients (N) Total observations (n)

174 267

GenderMale 70Female 78None specified 26

Age at observation2–17 years 3418–39 years 6140–59 years 121≥60 years 42None specified 9

Country of clinical studyCanada 3France 54Germany 19Hungary 7Italy/Belgium 35Japan 9Netherlands 31Switzerland 7United Kingdom 3United States 6

N: the number of individual LOPD patients. n: the number of observations usedin the analysis (maximum of 4 points per individual, with a minimum of 3months between measurements).

138 E.M. Johnson et al. /Neuromuscular Disorders 26 (2016) 136–145

MEP (not feasible; 0.08; 0.46), VC-U (0.30; 0.35; 0.24),FVC-U (0.78; 0.57; 0.47), VC-S (0.19; 0.23; 0.13), FVC-S (notfeasible; 0.48; not feasible), respectively. These p-values allexceed 0.05, suggesting that within each cohort there were nostatistical differences between the untreated and treated data.

3.2. Predictive thresholds for ventilator use

As shown in Table 2, non-invasive PFTs were able to describethe two predictive thresholds for ventilator use with high AUCvalues.The number of observations (n) used for each parameter arealso provided in this table. A visual representation of the datapoints and ROC curves for the 6 parameters (MIP, MEP, VC-U,VC-S, FVC-U, FVC-S) are shown in Fig. 2.

3.2.1. Threshold for nighttime ventilationIn differentiating the no ventilation and nighttime ventilation

groups, 5 metrics (MIP, FVC-U, VC-U, FVC-S, VC-S) wereable to predict these thresholds with AUC values ≥83%. For thetwo respiratory muscle strength metrics (MIP, MEP), only MIPwas able to describe the predictive threshold between the noventilation and nighttime ventilation groups with highdifferentiation (CP = 2.8 kPa, AUC = 89%, p < 0.0001). ForMEP, there was considerable overlap between the no ventilationand the nighttime ventilation groups, as characterized by amoderate AUC value (CP = 3.0 kPa, AUC = 72%, p = 0.03).

In examining the lung capacity tests, the cut points for VC-U(45% predicted) and FVC-U (39% predicted) for the thresholdsbetween the no ventilation and nighttime ventilation groups hadsimilar values. In the combined analysis of FVC-U and VC-U,the cut point (40% predicted) was between the values determinedfor the 2 individual metrics. For the supine vital capacitymeasurements, VC-S (CP = 25% predicted) provided gooddifferentiation between the no ventilator and nighttime ventilatorgroups. Unexpectedly, a lower AUC value was noted for FVC-S

which may reflect the relative lack of data reported for thisparameter. Supine measurements can be difficult to obtain inLOPD, as this position can result in severe dyspnea and may beintolerable. Consistent with the reported diaphragmatic weaknessin this disorder, the cut points for vital capacity in the supineposition were lower than the corresponding values determined inthe upright position.

3.2.2. Threshold for daytime ventilationIn determining the predictive thresholds between the nighttime

ventilation and daytime ventilation groups, five parameters (MIP,MEP, FVC-U, VC-U, VC-S) were, to varying degrees, able todifferentiate these two states with AUC values ≥80%, as shown inTable 2. There were not enough data points to evaluate thepredictive threshold for FVC-S. The predictive thresholds betweenthe nighttime and daytime cohorts for these parameters were alldistinctly lower than the values determined for the thresholdsbetween no ventilation and nighttime ventilation, indicating thatlung function decreased with disease progression.

For the respiratory muscle strength metrics, MIP describedboth ventilation thresholds with the same AUC values (89%). ForMEP, the ability to differentiate the two ventilator predicativethresholds was distinctly different, as shown in Table 2.

With the exception of FVC-S (there were too few data pointsfor this analysis), the lung capacity tests were able to predict thethreshold points between nighttime ventilation and daytimeventilatory assistance groups. Unexpectedly, the AUC values forFVC-U were lower for the threshold for daytime assistance(80%) than the nighttime value (87%). This decrease was notobserved with the VC-U values alone. However, in the combinedanalysis with both FVC-U and VC-U, the AUC values weredifferent for the thresholds between the no ventilation andnighttime groups (92%) and the nighttime and daytimeventilation groups (84%).

Table 2Prediction of nighttime and daytime ventilator use by non-invasive pulmonary function tests in LOPD.

Threshold for nighttime ventilator Threshold for daytime ventilator

Parameter None (n) Night (n) Cut point Sensitivity/specificity (%) AUC (%) Day (n) Cut point Sensitivity/specificity (%) AUC (%)

MIP (kPa) 37 43 2.8 63/95 89%p < 0.0001

21 1.1 57/98 89%p < 0.0001

MEP (kPa) 20 15 3.0 47/95 72%p = 0.03

18 2.4 89/80 95%p < 0.0001

FVC-U (% predicted) 47 37 39 35/98 87%p < 0.0001

28 17 32/97 80%p < 0.0001

VC-U (% predicted) 57 49 45 63/98 96%p < 0.0001

25 19 88/98 96%p < 0.0001

FVC-S (% predicted) 14 8 32 75/93 83%p = 0.01

– – – –

VC-S (% predicted) 23 19 25 58/96 96%p < 0.0001

14 12 71/95 86%p < 0.0004

VC-U & FVC-U(% predicted)

104 80 40 48/99 92%p < 0.0001

45 19 51/99 84%p < 0.0001

VC-S & FVC-S(% predicted)

36 27 33 81/94 94%p < 0.0001

14 12 71/89 89%p < 0.0001

Dash (–) indicates that there were not enough data points reported to perform this analysis. n: the number of observations assigned to each cohort. Cut point: valuewith the highest likelihood ratio from the ROC analysis. U: measurements performed in the upright position. S: measurements performed in the supine position. AUC(%): values of 100% represent a perfect discrimination between the 2 groups. None: group with no ventilation. Night: group with nighttime ventilation. Day: groupwith daytime ventilation.

139E.M. Johnson et al. /Neuromuscular Disorders 26 (2016) 136–145

Fig. 2. (A–F) Data points and receiver operating characteristic (ROC) curves for the 2 ventilator thresholds, no ventilation (none) – nighttime ventilation (night)groups and nighttime ventilation – daytime ventilation (day) groups. Mean (±SD) values are shown for each cohort as a line with error bars. (A) MIP (maximuminspiratory pressure, kPa). (B) MEP (maximum expiratory pressure, kPa). (C) FVC-U (forced vital capacity in the upright position, % predicted). (D) VC-U (vitalcapacity in the upright position, % predicted). (E) FVC-S (forced vital capacity in the supine position, % predicted). There were insufficient data points for patientswith daytime ventilation to complete the ROC analysis for the nighttime and daytime ventilation groups for FVC-S. (F) VC-S (vital capacity in the supine position,% predicted).

140 E.M. Johnson et al. /Neuromuscular Disorders 26 (2016) 136–145

To determine if there was a bias with the inclusion oflongitudinal data, an ROC analysis using only one data pointfrom each included patient was also performed. For individualswith multiple values, the lowest value was used in thisadditional analysis. For the two ventilator analyses (none andnight groups; night and day groups), the single data pointanalysis produced values for MIP (CP = 2.5 kPa, AUC = 89%;CP = 1.2 kPa, AUC = 90%), MEP (CP = 3.0 kPa, AUC = 67%;CP = 2.5 kPa, AUC = 92%), FVC-U (CP = 39% predicted,AUC = 86%; CP = 20% predicted, AUC = 80%), VC-U(CP = 45% predicted, AUC = 96%; CP = 18% predicted,AUC = 96%), FVC-S (CP = 31% predicted, AUC = 88%; noavailable data for daytime analysis), and VC-S (CP = 25%predicted, AUC = 96%; CP = 11% predicted, AUC = 89%),respectively. The similarities between these values and thecalculated values using the entire data set, as reported inTable 2, suggest that there was minimal/no bias with theinclusion of longitudinal data.

3.3. Predictive threshold for wheelchair use

To determine if measures of lung function could be used topredict the threshold between the mobile and wheelchairgroups, the identified data were grouped into 2 cohorts (mobile;wheelchair) and analyzed with ROC curves. As shown inTable 3, the pulmonary function parameters were able todifferentiate these two groups. Data points and ROC curves forthese analyses are also shown in Fig. 3A–F.

Among the PFTs, the respiratory muscle strength metrics(MIP, MEP) had the highest AUC values of 86% and 95%,respectively. With the exception of FVC-S (not enough data

points in the wheelchair cohort for an analysis), the lungcapacity metrics were also able to predict these thresholds. Incontrast with the ventilation analyses, the predictive thresholdsfor the vital capacity measurements for the upright and supinepositions were not distinctly different. In addition, the cut pointvalues were also quite low and well below the values observedfor the predictive threshold values determined for the noventilation to nighttime ventilatory support groups, supportinga discordance between progression of respiratory muscle vs.peripheral skeletal muscle weakness.

4. Discussion

The addition of ventilator assistance and wheelchair use aresignificant events in the progression of LOPD. In a study on thequality of life of 210 LOPD patients, the addition of ventilationand/or wheelchair use was associated with a maximal negativeimpact in the physical functioning scores on the SF-36 survey,with the these events ranked in the following order: ventilatorand wheelchair use > > ventilator only > wheelchair only [3].For these important events, clinically relevant surrogate tests toappropriately monitor disease progression would be beneficial.The tight associations observed between pulmonary functionmetrics and ventilator/wheelchair use suggest that lung functionparameters may be useful in monitoring the possible onset ofthese events.

Prior studies in patients with primary myopathies haveevaluated the relationship between respiratory function andpresence of sleep disordered breathing and diurnal respiratoryfailure. These studies demonstrated that as respiratory dysfunctionprogresses patients develop sleep disordered breathing whichprogresses to nocturnal hypoventilation, manifest as hypercapniaduring sleep, and eventually development of diurnal respiratoryfailure. The cut points identified in these prior studies for thepresence of nocturnal hypoventilation and diurnal respiratoryfailure coincides with the cut points identified in the present studyfor the nighttime and daytime ventilation groups, respectively. Ofimportance, these studies demonstrated that the onset of sleepdisordered breathing occurs at rather high levels of respiratoryfunction (vital capacity < 60% predicted, MIP < 4.5 kPa),indicating that the onset of sleep disordered breathing can be easilymissed without direct assessment of respiration and/or gasexchange during sleep. The present study extends theseobservations to a diverse group of patients with LOPD from manycountries and by identification of thresholds for supine vitalcapacity to account for the diaphragmatic weakness that ischaracteristic of LOPD.

In the present study, data for vital capacity were collectedfrom published studies that used differing measurementtechniques (e.g. FVC, slow VC, IVC). Whereas the measuredvital capacity will not vary by the respiratory maneuver inhealthy subjects, lower values for FVC can be observed inpatients with airway obstruction (e.g. asthma) [50]. In ourstudy, separate analyses were performed for FVC vs. VC.Although differences for the cut points were observed, theywere minimal and likely reflect variability in the underlyingpopulations used in these studies. Since, to the best of ourknowledge, there are no reports suggesting that respiratory

Table 3Prediction of wheelchair use by non-invasive pulmonary function tests inLOPD.

Parameter Threshold between mobile and wheelchair groups

Mobile(n)

Wheelchair(n)

Cutpoint

Sensitivity/specificity(%)

AUC (%)

MIP (kPa) 62 28 0.9 39/98 86%p < 0.0001

MEP (kPa) 28 22 2.6 90/96 95%p < 0.0001

FVC-U(% predicted)

81 22 13 23/99 84%p < 0.0001

VC-U(% predicted)

64 38 19 58/98 82%p < 0.0001

FVC-S(% predicted)

– – – – –

VC-S(% predicted)

13 17 12 59/92 74%p = 0.03

VC-U & FVC-U(% predicted)

142 49 13 22/93 83%p < 0.0001

VC-S & FVC-S(% predicted)

34 18 12 56/97 84%p < 0.0001

Dash (–) indicates that there were not enough data points reported to performthis analysis. n: the number of observations assigned to each cohort. Cut point:value with the highest likelihood ratio from the ROC analysis. U: measurementsperformed in the upright position. S: measurements performed in the supineposition. AUC (%): values of 100% represent a perfect discrimination betweenthe 2 groups.

141E.M. Johnson et al. /Neuromuscular Disorders 26 (2016) 136–145

dysfunction in LOPD is characterized by airway obstruction,the combined analysis which includes data from all studies mayprovide the most accurate assessment of the cut points.

Pulmonary function test criterion, in particular MIP and FVC,have been used as requirements for funding the equipment fornon-invasive ventilation. In the United States, the Centers forMedicare and Medicaid (CMS) requirements for fundingventilation assistance can be satisfied for neuromuscular disorderswith a criterion of either FVC < 50% predicted or MIP < 60cmH2O (5.9 kPa) [15]. These criteria were reportedly derived fromstudies in patients with amyotrophic lateral sclerosis (ALS)[51,52]. Even with these recommendations, Mendoza et al. (2007)found that patients with ALS reached the MIP criterion 4–6.5months earlier than the FVC criterion [52]. While our study did nothave sufficient data for a longitudinal analysis, our data for LOPD

patients did suggest that nocturnal and daytime mechanicalventilator support was more tightly linked to MIP values less than60 cmH2O (5.9 kPa) than to values of FVC or VC in the uprightposition below 50% predicted. Analysis of individual patientvalues revealed that 31% (17/55) of the LOPD patients usingnighttime ventilation had vital capacity measurements exceeding50% predicted, indicating that a large population of patients wouldnot qualify for ventilation reimbursement using either the CMSor NICE criteria in the absence of additional measurement ofrespiratory muscle strength. In contrast, only 8% (3/37) of theLOPD patients with nighttime ventilation had MIP valuesexceeding 60 cmH2O (5.9 kPa). Even for the more stringent NICEcriterion of MIP < 40 cmH2O (3.9 kPa), only 16% (6/37) ofthe LOPD patients did not meet these requirements. Inaddition, although all patients treated with nocturnal ventilation

Fig. 3. (A–F) Data points and receiver operating characteristic (ROC) curves for the thresholds between the mobile and wheelchair groups. (A) MIP (maximuminspiratory pressure, kPa). Mean (±SD) values are shown for each cohort as a line with error bars. (B) MEP (maximum expiratory pressure, kPa). (C) FVC-U (forcedvital capacity in the upright position, % predicted). (D) VC-U (vital capacity in the upright position, % predicted). (E) FVC-S (forced vital capacity in the supineposition, % predicted). (F) VC-S (vital capacity in the supine position, % predicted).

142 E.M. Johnson et al. /Neuromuscular Disorders 26 (2016) 136–145

demonstrated values for VC or FVC in the supine position below50% predicted, it is unclear if these supine measurementswould satisfy either the CMS or NICE guidelines, especially ifsimultaneous upright values were above this threshold. With theseguidelines for ventilation reimbursement, these data demonstratethat many subjects will achieve the MIP criteria before the VCcriteria.

The predictive threshold for VC for nighttime ventilation wassimilar to the criteria that justifies reimbursement by CMS,suggesting a potential bias. However, this study included patientsfrom multiple countries, and reimbursement thresholds wereonly in place for the US and recently (since 2010) for the UK(National Institute for Health and Care Excellence, NICE). Therewere only a total of 9 individuals in this study who were fromeither the US or the UK, as shown in Table 1. While notexcluding the possibility that physicians in the other countriescould have used the CMS/NICE guidelines, reimbursement ofNIV therapy in the majority of patients included in our studywould not have been contingent on the measured lung function.Moreover, approximately 1/3 of the patients receiving nocturnalventilation had measured values of vital capacity exceeding 50%predicted, indicating that other criteria for initiating therapy hadclearly been used. It is also unlikely that MIP criteria were usedin all of these subjects, as this parameter was monitored lessfrequently in the patients included in our study. Although theretrospective nature of our study does not allow definitivedetermination of the indication for nocturnal ventilation inindividual patients, the thresholds that we determined werehighly reproducible despite variable locations/countries andprescribing physicians. Lastly, the thresholds we identified fornocturnal ventilation are physiologically plausible, as thesevalues coincide with the level of respiratory function wherehypoventilation first manifests during sleep in patients withrespiratory muscle weakness, as discussed previously.

As a heterogeneous disorder, the progressive muscleinvolvement in LOPD can vary substantially with diverserequirements for mobility and respiratory assistive devices[53]. While wheelchair assistance is generally thought to berelated to decreased function of the motor muscles, there maybe a respiratory or endurance component that contributes tothe decision to use a wheelchair. Case study observations havesuggested that respiratory distress or exertional dyspnea maybe factors in wheelchair use [21,22]. Our analysis does indicatethat PFTs (MIP, MEP, FVC-U, VC-U, VC-S) can be used todistinguish mobility from wheelchair use. In particular, ouranalysis suggested that the threshold for wheelchair use coincidedwith the need for daytime ventilation assistance. Althoughspeculative in nature, the use of predictive thresholds forwheelchair use is a novel finding, which warrants additionalinvestigation. We cannot exclude, for example, a direct influenceof the seated position and/or scoliosis in the evolution ofrespiratory status (e.g. the lack of exercise may, by itself,decrease pulmonary function). In addition, the equipmentrequirements for daytime ventilation may influence the use ofa wheelchair for mobility. Prospective, longitudinal studies toevaluate the multiple factors that may influence wheelchair usewould be needed to fully characterize this mobility transition.

Our study may be limited by the heterogeneity of thereported data. Different methodologies for determining the PFTsmay contribute to variability in these analyses. In addition, as across-sectional study, the individual requirements for initiatingventilation and wheelchair use cannot be discerned. Longitudinalstudies monitoring PFTs, SDB, nocturnal hypoventilation, anddiurnal respiratory failure, as well as ventilation and wheelchairevents, would be beneficial. Previous studies have also shown astrong association between pulmonary function and scoliosis inLOPD patients [54]. In a Pompe registry, scoliosis was reportedin the majority (62.5%) of wheelchair users for all age groups.Also, a larger percentage of patients with scoliosis requiredrespiratory support than patients without scoliosis (44% vs.27.2%, respectively). In the included studies, data describing thepresence and degree of scoliosis were uncommon, precluding anassessment of the impact of scoliosis on the entire study. It shouldbe noted, however, that scoliosis may contribute to both theobserved reduced pulmonary function and wheelchair use.

5. Conclusion

In LOPD, the addition of ventilation and/or wheelchair use aresignificant events in disease progression, negatively affectingquality of life. Our findings suggest that measures of pulmonaryfunction (MIP, MEP, FVC, VC) can act as surrogate markers topredict the thresholds for ventilation (nighttime; daytime). MIPand VC in the upright position had high AUC values andconsistency in predicting these thresholds. In comparison to sleepstudies reported in the literature, the threshold points for nighttimeventilation coincided with values that are associated with presenceof nocturnal hypoventilation in primary myopathies. In anexploratory analysis on a possible relationship between respiratoryfunction and mobility, pulmonary function tests were alsoassociated with wheelchair use. Pulmonary function tests may aidin identifying patients at risk for ventilation and monitoring diseaseprogression in patients with late-onset Pompe disease.

Acknowledgements

This study was supported by BioMarin Pharmaceutical Inc.Johnson and Quartel are employees and stockholders ofBioMarin. All other investigators provided consulting services,received research grants, participated in advisory boardmeetings, received speaker honoraria, and/or were provided withtravel support from BioMarin Pharmaceutical Inc.

References

[1] Winkel LP, Hagemans ML, van Doorn PA, et al. The natural course ofnon-classic Pompe’s disease; a review of 225 published cases. J Neurol2005;252:875–84.

[2] Gungor D, de Vries JM, Hop WC, et al. Survival and associated factors in268 adults with Pompe disease prior to treatment with enzymereplacement therapy. Orphanet J Rare Dis 2011;6:34.

[3] Hagemans M, Janssens A, Winkel L, et al. Late-onset Pompe diseaseprimarily affects quality of life in physical health domains. Neurology2004;63:1688–92.

[4] Hagemans ML, Laforet P, Hop WJ, et al. Impact of late-onset Pompedisease on participation in daily life activities: evaluation of the RotterdamHandicap Scale. Neuromuscular Disord 2007;17:537–43.

143E.M. Johnson et al. /Neuromuscular Disorders 26 (2016) 136–145

[5] Wokke JH, Escolar DM, Pestronk A, et al. Clinical features of late-onsetPompe disease: a prospective cohort study. Muscle Nerve 2008;38:1236–45.

[6] Mellies U, Stehling F, Dohna-Schwake C, Ragette R, Teschler H, Voit T.Respiratory failure in Pompe disease: treatment with noninvasiveventilation. Neurology 2005;64:1465–7.

[7] Boentert M, Karabul N, Wenninger S, et al. Sleep-related symptoms andsleep-disordered breathing in adult Pompe disease. Eur J Neurol 2015;22:369–76, e27.

[8] Mellies U, Ragette R, Schwake C, Boehm H, Voit T, Teschler H. Daytimepredictors of sleep disordered breathing in children and adolescents withneuromuscular disorders. Neuromuscular Disord 2003;13:123–8.

[9] Mellies U, Ragette R, Schwake C, Baethmann M, Voit T, Teschler H.Sleep-disordered breathing and respiratory failure in acid maltasedeficiency. Neurology 2001;57:1290–5.

[10] Ragette R, Mellies U, Schwake C, Voit T, Teschler H. Patterns andpredictors of sleep disordered breathing in primary myopathies. Thorax2002;57:724–8.

[11] Toussaint M, Steens M, Soudon P. Lung function accurately predictshypercapnia in patients with Duchenne muscular dystrophy. Chest Journal2007;131:368–75.

[12] Young H, Lowe A, Fitzgerald D, et al. Outcome of noninvasive ventilationin children with neuromuscular disease. Neurology 2007;68:198–201.

[13] Ward S, Chatwin M, Heather S, Simonds A. Randomised controlled trialof non-invasive ventilation (NIV) for nocturnal hypoventilation inneuromuscular and chest wall disease patients with daytime normocapnia.Thorax 2005;60:1019–24.

[14] Dreher M, Rauter I, Storre JH, Geiseler J, Windisch W. When shouldhome mechanical ventilation be started in patients with differentneuromuscular disorders? Respirology 2007;12:749–53.

[15] National Government Services I, Local Coverage Determination (LCD):respiratory assist devices (L27228), Services C f M a M, Editor.

[16] Excellence NIfHaC, Motor neurone disease: the use of non-invasiveventilation in the management of motor neurone disease, 2010.

[17] de Vries JM, Brugma JD, Ozkan L, et al. First experience with enzymereplacement therapy during pregnancy and lactation in Pompe disease.Mol Genet Metab 2011;104:552–5.

[18] Demey HE, Van Meerbeeck JP, Vandewoude MF, Prove AM, Martin JJ,Bossaert LL. Respiratory insufficiency in acid maltase deficiency: theeffect of high protein diet. JPEN 1989;13:321–3.

[19] Deroma L, Guerra M, Sechi A, et al. Enzyme replacement therapy injuvenile glycogenosis type II: a longitudinal study. Eur J Pediatr2014;173:805–13.

[20] Diab K, Snook RJ, Farber M. A 46-year-old man with dyspnea and leftdiaphragm paralysis. Chest 2009;136:1690–3.

[21] Furusawa Y, Mitsuhashi S, Mori-Yoshimura M, et al. Late-onset Pompedisease after 4 years of enzyme replacement therapy: an autopsy case.Neurol Clin Neurosci 2014;2:7–9.

[22] Furusawa Y, Mori-Yoshimura M, Yamamoto T, et al. Effects of enzymereplacement therapy on five patients with advanced late-onset glycogenstorage disease type II: a 2-year follow-up study. J Inherit Metab Dis2012;35:301–10.

[23] Gaeta M, Barca E, Ruggeri P, et al. Late-onset Pompe disease (LOPD):correlations between respiratory muscles CT and MRI features andpulmonary function. Mol Genet Metab 2013;110:290–6.

[24] Hobson-Webb LD, Jones HN, Kishnani PS. Oropharyngeal dysphagiamay occur in late-onset Pompe disease, implicating bulbar muscleinvolvement. Neuromuscular Disord 2013;23:319–23.

[25] Hundsberger T, Rosler KM, Findling O. Cessation and resuming ofalglucosidase alfa in Pompe disease: a retrospective analysis. J Neurol2014;261:1684–90.

[26] Illes Z, Mike A, Trauninger A, Vardi K, Vaczi M. Motor function andrespiratory capacity in patients with late-onset pompe disease. MuscleNerve 2014;49:603–6.

[27] Jones HN, Moss T, Edwards L, Kishnani PS. Increased inspiratory andexpiratory muscle strength following respiratory muscle strength training(RMST) in two patients with late-onset Pompe disease. Mol Genet Metab2011;104:417–20.

[28] Kawamoto M, Ishii J, Togo M, et al. P173: seven-year follow up of twosisters with late onset Pompe’s disease: effects and limitation of enzymereplacement therapy. Clin Neurophysiol 2014;125:S93–4.

[29] Khan A, McNeil C, Casey R Improved or stabilized respiratory function in3 patients with Pompe disease after starting recombinant alpha-glucosidase(rhGAA)-a case to maintain a steady course with treatment. Clin Therap2010;32:S77.

[30] Lilleker J, Roberts M, Boothman B. Dilative vasculopathy and cerebralhaemorrhage as a presentation of late–onset pompe disease. J NeurolNeurosurg Psychiatry 2013;84:e2.

[31] Martin RJ, Sufit RL, Ringel SP, Hudgel DW, Hill PL. Respiratoryimprovement by muscle training in adult-onset acid maltase deficiency.Muscle Nerve 1983;6:201–3.

[32] Merk T, Wibmer T, Schumann C, Kruger S. Glycogen storage diseasetype II (Pompe disease)–influence of enzyme replacement therapy inadults. Eur J Neurol 2009;16:274–7.

[33] Nabatame S, Taniike M, Sakai N, et al. Sleep disordered breathing inchildhood-onset acid maltase deficiency. Brain & Development 2009;31:234–9.

[34] Orlikowski D, Pellegrini N, Prigent H, et al. Recombinant human acidalpha-glucosidase (rhGAA) in adult patients with severe respiratoryfailure due to Pompe disease. Neuromuscular Disord 2011;21:477–82.

[35] Papadimas GK, Spengos K, Konstantinopoulou A, et al. Adult Pompedisease: clinical manifestations and outcome of the first Greek patientsreceiving enzyme replacement therapy. Clin Neurol Neurosurg 2011;113:303–7.

[36] Pellegrini N, Laforet P, Orlikowski D, et al. Respiratory insufficiency andlimb muscle weakness in adults with Pompe’s disease. Eur Respir J2005;26:1024–31.

[37] Remiche G, Herbaut A, Ronchi D, et al. Incontinence in Late OnsetPompe disease: an underdiagnosed although potentially treatablecondition. In 62nd Annual Meeting of American Academy of Neurology.2011.

[38] Schneider I, Hanisch F, Muller T, Schmidt B, Zierz S. Respiratoryfunction in late-onset Pompe disease patients receiving long-term enzymereplacement therapy for more than 48 months. Wiener MedizinischeWochenschrift 2013;163:40–4.

[39] Schneider I, Hanisch F, Müller T, Zierz S. Respiratory function in lateonset Pompe disease patients upon long-term enzyme replacementtherapy. Klinische Neurophysiologie 2012;43:P105.

[40] Sivak ED, Ahmad M, Hanson MR, Mitsumoto H, Wilbourn AJ.Respiratory insufficiency in adult-onset acid maltase deficiency. SouthMed J 1987;80:205–8.

[41] Swash M, Schwartz M, Apps M. Adult onset acid maltase deficiency:distribution and progression of clinical and pathological abnormality in afamily. J Neurol Sci 1985;68:61–74.

[42] van Capelle CI, Winkel LP, Hagemans ML, et al. Eight years experiencewith enzyme replacement therapy in two children and one adult withPompe disease. Neuromuscular Disord 2008;18:447–52.

[43] van der Beek NA, Hagemans ML, Reuser AJ, et al. Rate of diseaseprogression during long-term follow-up of patients with late-onset Pompedisease. Neuromuscular Disord 2009;19:113–17.

[44] van der Beek NA, van Capelle CI, van der Velden-van Etten KI, et al. Rateof progression and predictive factors for pulmonary outcome in childrenand adults with Pompe disease. Mol Genet Metab 2011;104:129–36.

[45] Vianello A, Semplicini C, Paladini L, et al. Enzyme replacement therapyimproves respiratory outcomes in patients with late-onset type IIglycogenosis and high ventilator dependency. Lung 2013;191:537–44.

[46] Winkel LP, Van den Hout JM, Kamphoven JH, et al. Enzyme replacementtherapy in late-onset Pompe’s disease: a three-year follow-up. Ann Neurol2004;55:495–502.

[47] Carlier RY, Laforet P, Wary C, et al. Whole-body muscle MRI in 20patients suffering from late onset Pompe disease: involvement patterns.Neuromuscular Disord 2011;21:791–9.

[48] Mundy H, Williams J, Cousins A, Lee P. The effect of L-alanine therapyin a patient with adult onset glycogen storage disease type II. J InheritMetab Dis 2006;29:226–9.

144 E.M. Johnson et al. /Neuromuscular Disorders 26 (2016) 136–145

[49] Vielhaber S, Brejova A, Debska-Vielhaber G, et al. 24-months results intwo adults with Pompe disease on enzyme replacement therapy. ClinNeurol Neurosurg 2011;113:350–7.

[50] Chhabra S. Forced vital capacity, slow vital capacity, or inspiratory vitalcapacity: which is the best measure of vital capacity? J Asthma1998;35:361–5.

[51] Miller RG, Rosenberg JA, Gelinas D, et al. Practice parameter: the care ofthe patient with amyotrophic lateral sclerosis (an evidence-based review)Report of the Quality Standards Subcommittee of the American Academyof Neurology. Neurology 1999;52:1311.

[52] Mendoza M, Gelinas DF, Moore DH, Miller RG. A comparison ofmaximal inspiratory pressure and forced vital capacity as potential criteriafor initiating non-invasive ventilation in amyotrophic lateral sclerosis.Amyotroph Lateral Scler 2007;8:106–11.

[53] Hagemans M, Winkel L, Van Doorn P, et al. Clinical manifestation andnatural course of late-onset Pompe’s disease in 54 Dutch patients. Brain2005;128:671–7.

[54] Roberts M, Kishnani PS, van der Ploeg AT, et al. The prevalence andimpact of scoliosis in Pompe disease: lessons learned from the PompeRegistry. Mol Genet Metab 2011;104:574–82.

145E.M. Johnson et al. /Neuromuscular Disorders 26 (2016) 136–145