1

VENTILATION HETEROGENEITY IN THE ACINAR AND CONDUCTIVE 1

ZONES OF THE NORMAL AGEING LUNG 2

3

Sylvia Verbanck 1, Bruce R. Thompson 2, Daniel Schuermans 1, Harpal S. Kalsi 3, 4

Martyn F. Biddiscombe 3, Chris Stuart-Andrews 2, Shane Hanon 1, Alain Van Muylem 4, 5

Manuel Paiva 4 , Walter Vincken 1 and Omar S. Usmani 3. 6

7

1 Respiratory Division, University Hospital UZ Brussel, 1090 Brussels, Belgium 8 2 Allergy, Immunology and Respiratory Medicine, The Alfred Hospital, 3004 Melbourne, 9 Australia. 10 3 National Heart and Lung Institute, Imperial College London and Royal Brompton 11 Hospital, London, United Kingdom. 12 4 Respiratory Division, University Hospital Erasme, Université Libre de Bruxelles, 1070 13 Brussels, Belgium 14 15 16 17 Corresponding author : Sylvia Verbanck 18

Respiratory Division, University Hospital UZ Brussel 19 Laarbeeklaan 101 20 1090 Brussels, Belgium 21 Tel : +32 2 477 6346 22 Fax : +32 2 477 6352 23

Email: sylvia.verbanck@uzbrussel.be 24 25 26 27 28 running head: ventilation heterogeneity in the ageing lung. 29 30 Word count manuscript body : 3096 31 32 33 The Corresponding Author has the right to grant on behalf of all authors and does grant on behalf of all 34 authors, an exclusive licence (or non exclusive for government employees) on a worldwide basis to the 35 BMJ Publishing Group Ltd and its Licensees to permit this article (if accepted) to be published in Thorax 36 editions and any other BMJPGL products to exploit all subsidiary rights, as set out in our 37 licence(http://group.bmj.com/products/journals/instructions-for-authors/licence-forms/). 38 39 40 Competing Interest: None to declare.41

2

ABSTRACT (241 words) 42 43

Rationale: Small airways function studies in lung disease have used 3 promising 44

multiple breath washout (MBW) derived indices: Indices of ventilation heterogeneity in 45

the acinar (Sacin) and conductive (Scond) lung zones, and the lung clearance index (LCI). 46

Since peripheral lung structure is known to change with age, ventilation heterogeneity is 47

expected to be affected too. However, the age-dependence of the MBW indices of 48

ventilation heterogeneity in the normal lung is unknown. 49

Objectives: We systematically investigated Sacin, Scond or LCI as a function of age, 50

testing also the robustness of these relationships across two laboratories. 51

Methods: MBW tests were performed by never-smokers (50% male) in the age range 52

25-65 years, with data gathered across two laboratories (n=120 and n=60). For 53

comparison with the literature, the phaseIII slopes from classical single breath washout 54

(SBW) tests were also acquired in one group (n=120). 55

Measurements and Main Results: All three MBW indices consistently increased with 56

age, representing a steady worsening of ventilation heterogeneity in the age range 25-57

65. Age explained 7-16% of the variability in Sacin and Scond and 36% of the 58

variability in LCI. There was a small but significant gender difference only for Sacin. 59

Classical SBW phaseIII slopes also showed age-dependencies, with gender effects 60

depending on the normalization method used. 61

Conclusions: With respect to the clinical response, age is a small but consistent effect 62

that needs to be factored in when using the MBW indices for the detection of small 63

airways abnormality in disease. 64

65

keywords: small airways function, acinar airways, conductive airways, ventilation 66

distribution, multiple breath washout. 67

68

3

69

what is the key question : It has been shown that alveolar architecture changes with 70

age, however the age dependence of lung function indices that can actually measure 71

functional change in the alveolar region have yet to be investigated. 72

what is the bottom line : If the alveolar architecture changes with age are large 73

enough to also be reflected functionally in small airway indices, these need to be 74

acknowledged, as early changes in the small airways may in fact be a normal ageing 75

effect. 76

why read on : While previous studies clearly show clinical usefulness of indices that 77

reflect gas mixing within the small airways, the present study illustrates how neglecting 78

the effect of age can unduly lead to diagnosis of small airway dysfunction in older 79

subjects. 80

81

82

4

INTRODUCTION 83

84

The multiple breath washout (MBW) test has been advocated for small airways 85

detection in obstructive lung disease.[1,2] Initially propelled by paediatric clinical lung 86

research but now also promoted in adult lung disease, the most frequently reported 87

MBW index is the lung clearance index (LCI).[3-5] First introduced in 1951 as a 88

measure of overall lung ventilation heterogeneity,[6] LCI is currently deemed useful in 89

the early detection of cystic fibrosis lung disease,[3] with the specific advantages that 90

LCI requires no particular breathing volumes and is quasi independent of lung growth 91

(1-17years).[7] In adults, an analysis of the MBW phaseIII slope was proposed to 92

distinguish ventilation heterogeneity generated in acinar air spaces from that originating 93

in the more proximal lung.[8] The two most relevant MBW derived indices of ventilation 94

heterogeneity are referred to as Sacin (for ventilation heterogeneity generated 95

peripheral to the acinar entrance) and Scond (for ventilation heterogeneity generated in 96

the conductive lung zone). An exhaustive review of theory and experiments 97

underpinning this phaseIII slope analysis are part of a recent update of the Handbook of 98

Physiology,[9] and the most critical aspects are iterated in the online supplement. Due 99

to intrinsic structural heterogeneity of the airways within the lungs, Sacin and Scond are 100

non-zero in the normal lung, but show marked increases in lung disease.[10-16] In 101

patients with chronic obstructive pulmonary disease (COPD) those with emphysema 102

show a greater Sacin;[10] the extent of Sacin increase is associated with carbon 103

monoxide transfer factor,[11] and with high resolution computed-tomography lung 104

density.[12] In asthma, patients with an increased Sacin are better responders to small 105

particle-sized corticosteroids,[13] and Sacin correlates with an increased alveolar nitric 106

oxide.[14] In adult asthma, Scond is a predictor of airway hyperresponsiveness, 107

5

independent of inflammation,[15] and in preschool wheezers Scond is a sensitive 108

indicator of abnormal pulmonary function.[16] 109

110

To date, no comprehensive reports exist of LCI, Sacin and Scond values and 111

their dependence on age in a normal adult population. On the one hand, early 112

physiological studies using the phaseIII slope of the single breath washout (SBW) 113

following a one-litre oxygen inhalation, have indicated an age dependence of ventilation 114

heterogeneity.[17,18] In later studies, age dependence of the vital capacity SBW 115

phaseIII slope – which includes a gravitational component,[19] – was found to be either 116

non-existent,[20] or poor up to the age of 55-60 years.[21-23] On the other hand, 117

histological evidence,[24] and indirect measurement by magnetic resonance imaging,[ 118

25] have demonstrated alveolar size increases in the normal lung between 25-65years. 119

This is expected to impact on ventilation heterogeneity in the most peripheral acinar 120

lung units (potentially affecting Sacin) but also on the elastic properties of clusters of 121

acini (potentially affecting Scond). Given the potential associations between LCI and 122

Sacin or Scond,[26,27] LCI is also likely to be affected in a normal ageing adult lung, as 123

opposed to age-independence of LCI in children.[7] We therefore hypothesized that 124

while Sacin is the most likely MBW index to be age-dependent, Scond and LCI may be 125

affected by age as well. 126

127

128

MATERIALS AND METHODS 129

130

MBW tests were collected on never-smoker healthy subjects in the age range 25-131

65 years, in two laboratories (n=120 and n=60). In order to test the potential for 132

automated analysis of such large MBW data sets for future clinical use, similar to that 133

6

previously done for the SBW test,[28] we also submitted a data subset for semi-134

automated analysis recently developed by a third participating laboratory.[29] The study 135

protocols at UZ Brussel (core dataset site) and Brompton Hospital (supplemental 136

dataset site) were approved by the respective local research ethics committee 137

(#B14320097554; 08/H0709/2). All participating subjects were Caucasian, were not 138

obese (defined as a body mass index >30) and were defined as healthy through clinical 139

screening with (i) an absence of history of symptoms suggestive of respiratory disease, 140

(ii) no childhood or past medical history of respiratory disease, and (iii) had never 141

smoked. Subject recruitment was undertaken through open advertisement in an 142

intention-to-enter manner to avoid potential bias or lack of representativeness of the 143

population at large. All subjects provided written informed consent prior to testing. 144

145

Core dataset (UZ Brussel; n=120). 146

After standard spirometry, MBW and SBW tests were performed in triplicate on 147

120 subjects (15M/15F in each decade between 25-65 years of age). The MBW test 148

involved one-litre tidal breathing from functional residual capacity (FRC). The SBW test 149

was performed as two previously used maneuvers,[17,18,20-22] : either a one-litre 150

inspiration from FRC with expiration to residual volume (SBWFRC) or a vital capacity 151

inspiration with expiration to back to residual volume (SBWRV). A bag-in-box and valve 152

system was used, with a re-inspired dead space (i.e., re-inspired volume in subsequent 153

MBW inhalations) amounting to 50ml, and a N2 analyzer (PK-Morgan,UK). 154

155

Supplemental dataset (Brompton Hospital; n=60). 156

After spirometry, MBW tests were performed in triplicate on 60 subjects between 157

25-65yrs, with an even spread of gender across the age range studied. The equipment 158

7

was very similar to the UZ Brussel bag-in-box setup, with a re-inspired dead space of 159

15ml, and a N2 analyzer (Logan Research, Kent, UK). 160

161

After accounting for synchronization between N2 and pneumotachograph signals 162

and re-inspired dead space specific to core and supplemental data sets, MBW tests 163

were analyzed as per instructions provided by co-author (MP) common to the present 164

and the original paper on MBW analysis.[8] To this end, custom-built manually 165

operated software was used by authors SV and MP to determine phaseIII slopes 166

(nominally between 0.65 and 1L) from each breath of the MBW test. The phaseIII 167

slopes were normalized by the mean expired N2 concentration, plotted against lung 168

turnover, and Sacin and Scond were calculated after pooling all valid normalized slopes 169

from the 3 MBW tests;[9] a typical example is shown in the online supplement. 170

Additionally, LCI was computed as the number of lung turnovers where mean expired 171

N2 concentration had reached 1/40th of the pre-test alveolar N2 concentration; this was 172

also done on the average of 3 N2 concentration washout curves (with each washout 173

curve normalized to its pre-test alveolar N2 concentration); the rationale and impact of 174

using mean expired or end-tidal concentration for LCI computation is discussed 175

elsewhere [27]. From the SBW tests (pertaining to the core data set), phaseIII slope 176

was determined over the entire expiration down to residual volume, and normalized by 177

mean expired concentration. Depending on the SBW starting volume (FRC or RV), the 178

normalized phaseIII slope is referred to as SnIII_FRC or SnIII_RV. For the SBW 179

manuever starting from RV, the phaseIII slope is also reported as a non-normalized 180

phaseIII slope (SIII_RV; in %/L) for ease of comparison with the literature.[20,23] 181

182

183

Statistical analysis 184

8

All statistical analyses were performed using MedCalc (v10.4, Mariakerke, 185

Belgium). Variables were tested for normality using the Chi-squared test. Multiple 186

stepwise regressions were performed on Sacin, Scond, LCI, SnIII_FRC 187

(logtransformed), SnIII_RV, and SIII_RV (logtransformed), including gender, age, 188

height, and functional residual volume as independent variables. A forward regression 189

was used, setting the inclusion criterion for retention at P<0.05. When gender was 190

retained as a significant factor, the multiple regression analysis was repeated for each 191

gender. Plots of standardised residuals against standardised predicted values were 192

visually inspected for the validity of linearity and homoscedacity. Analysis of covariance 193

was used to test for differences in Sacin, Scond and LCI between the datasets from the 194

two laboratories (factor: laboratory; covariate: age). Statistical significance was 195

accepted at P<0.05. Bland Altman plots were used to examine agreement between 196

manual MBW analysis by experts and semi-automated MBW analysis. 197

198

199

RESULTS 200

201

Summary data from the core and supplemental datasets are given in Table 1; the 202

table also includes SBW data from the core data set. Figures 1-3 show the Sacin, 203

Scond and LCI increase as a function of age. In the multiple regression analyses on 204

Sacin, Scond and LCI values of the core data set, height or FRC did not reach statistical 205

significance for inclusion in the regression model, while age was a significant contributor 206

to all three MBW indices (P<0.001); Sacin was the only index showing a significant 207

gender-dependence (P=0.002). Age-dependent regression equations corresponding to 208

the core dataset can be found in Table 2, with gender specific equations only for Sacin. 209

When dividing the regression slopes by the value obtained for a subject corresponding 210

9

to the middle of the age range under study (45yr), these relative increases of Sacin, 211

Scond and LCI were ranged 0.4 - 1.2 % per year of increasing age. For comparison to 212

SBW data in the literature, regression equations are also included for SBW derived 213

phaseIII slopes (Table 3). In addition to age, both height and FRC were independent 214

predictors depending on the SBW maneuver and phaseIII slope normalization. In fact, 215

in the case of the vital capacity SBW, the phaseIII normalization method determined 216

whether the slope was different between both genders or not (SnIII_RV versus 217

SIII_RV). 218

219

The panels B of Figures 1-3 compare the core dataset to the supplemental 220

dataset for Sacin, Scond and LCI. Analysis of covariance showed no difference 221

between laboratories for Sacin and Scond, whereas LCI was significantly lower in the 222

supplementary versus the core data set (mean LCI difference: -0.54; P<0.0001). The 223

prediction of why such a difference should be more pronounced for LCI than for Sacin 224

or Scond,[30] and some additional experiments to verify this, can be found in the online 225

supplement. Also included in an online supplement are the specifications of an 226

automated MBW analysis, applying a recently proposed break point algorithm.[29] 227

Using this technique, we re-analyzed half the core data set (in order of subject entry) 228

and the resulting Bland-Altman plots in Figure 4 show that, for the most difficult 229

parameters to obtain automatically (Sacin and Scond), a good agreement between 230

automated and manual analysis was obtained. 231

232

233

DISCUSSION 234

235

10

While normal ageing has been shown to be associated with decreased lung 236

compliance and decline in airway diameter,[31] little is known about intrapulmonary 237

heterogeneity of these changes with age. Since ventilation heterogeneity involves 238

dynamic processes, and cannot be reduced to static morphometric properties such as 239

airway wall thickness, physiological tests of ventilation heterogeneity are the only way to 240

investigate this. In the present study, we have quantified the age-dependent change of 241

ventilation heterogeneity in different lung zones via the MBW test, demonstrating a 242

significant age dependence on Sacin, Scond and LCI in healthy never-smoker subjects. 243

Height was shown not to have a significant impact on any of the 3 MBW indices under 244

study, and gender affected only Sacin. The age-dependence of Sacin was very similar 245

across both gender subgroups (Figure1A) and the largest absolute Sacin difference 246

between males and females (estimated at 65yrs (Table 2): 0.025L-1) was small with 247

respect to values encountered in disease where Sacin typically amount to 0.2 L-1 in 248

moderate asthma,[13] and rises to 0.3-0.4 L-1 in COPD patients depending on the 249

presence of emphysema or not.[10] Nevertheless, when setting up clinical studies for 250

early detection purposes, the gender dependence of Sacin may need to be taken into 251

account. 252

253

The slightly greater Sacin in male subjects could have suggested that the intra-254

acinar bifurcation pattern, i.e., a major determinant of acinar ventilation heterogeneity, is 255

more asymmetric in men. However, we contend that the observed M/F difference in 256

Sacin is more likely due to the fact that inherent to MBW, Sacin is based on a 257

normalized phaseIII slope from an exhalation that has been truncated at 1L instead of 258

exhalation to residual volume. This can be inferred from comparison of Sacin with the 259

SBW derived SnIII_FRC, which is computed from an exhalation that continues to 260

residual volume, such that all acinar ventilation heterogeneities can fully contribute to 261

11

the phase III, i.e., to its slope and to the mean expired concentration (by which the 262

phaseIII slope is to be normalized). When dividing the maximum Sacin difference 263

between M and F (occurring at 65years : 0.025L-1) by its average Sacin value (0.111L-1) 264

predicted from the equations in Table 2, we obtain a relative M//F difference of 22%. 265

The corresponding number based on the prediction equations in Table 3 for SnIII_FRC 266

amounts to only 3% (in this case, average heights and weights from Table 1 are used to 267

compute the predicted values M and F of SnIII_FRC). At lower ages the dependence of 268

Sacin and SnIII_FRC on gender is even reversed, depending also on the choice of FRC 269

and height in the case of SnIII_FRC. Hence, we conclude that the gender dependence 270

of Sacin, which is not negligible with respect to the low Sacin values in normal subjects 271

is an intrinsic feature of MBW normalized phaseIII slope analysis and a trade-off to 272

enable the distinction between acinar and conductive ventilation heterogeneities in lung 273

disease. 274

275

In fact, distinct gender differences were observed in the very first physiological 276

investigations of the age-dependence of ventilation distribution,[17,18] using the 277

phaseIII slope derived from the SBW maneuver starting from FRC. While this was in 278

large part due to absence of phaseIII slope normalization, later studies using the vital 279

capacity SBW maneuver also showed a gender difference in phaseIII slope.[20,23] For 280

this SBW maneuver, the SIII_RV difference is opposite to that for Sacin, essentially due 281

to a slightly greater RV/TLC in females versus males.[20,23] For a 45yr old female vs 282

male, predicted SIII_RV is 1.21(F) vs 0.96(M) %/L in the present study, 1.40 vs 1.04 283

%/L in the Portland cohort in Buist et al,[20] and 1.11 vs 0.88%/L in a London cohort in 284

Roberts et al,[23]. When normalizing the vital capacity SBW phaseIII slope by mean 285

expired concentration in the present study, the gender gap disappears altogether 286

(SnIII_RV in Table 3). 287

12

288

The LCI dependence on age was consistent, despite an absolute LCI difference 289

between both participating laboratories associated with an equipment dead space (i.e., 290

re-inspired dead space) difference. Since the value of this particular dead space 291

essentially depends on the valve system, this cannot be realistically imposed on all 292

laboratories wanting to incorporate MBW tests to their lung function equipment. Despite 293

the many potential equipment and methodological differences between our study and 294

that by Bouhuys in 1963,[32] the lung clearance index they measured (in males 24-295

65yrs) showed an increase between the average value reported for the lowest and 296

highest decade (9.1 vs 10.0). The increase of LCI with age found here in adults, is in 297

contrast to what was previously reported in children and adults.[7,26] The observed 298

0.22 unit LCI change per decade found here is also not negligible with respect to what is 299

thought to be a clinically relevant LCI difference of typically 0.45 between a healthy 300

control and asthma group.[4] Clearly, when studying ventilation heterogeneity in terms 301

of LCI in adults, age is a small but consistent factor that needs to be acknowledged. 302

303

The relative increases of the three MBW indices with age are small, ranging 0.4% 304

(for LCI) to 1.2% (for Sacin), yet, the actual impact of age on LCI or on Sacin can also 305

be illustrated as follows. For a 25 year old subject, the LCI upper limit of normal 306

(computed as the predicted value +1.645*RSD) would be 6.39, which corresponds to 307

the predicted value of a normal 50 year old. With a similar reasoning, the Sacin and 308

Scond upper limit of normal for a 25 year old subject would correspond to the respective 309

predicted values for subjects older than 65 years. Together with the greater R2 value 310

for LCI, this example illustrates that it is even more important for LCI than for Sacin or 311

Scond to factor in age when using these MBW indices as diagnostic tools. Absolute 312

values for the upper limits of normal for LCI, Sacin or Scond may differ between 313

13

laboratories, depending on instrumental dead space (expected to affect primarily LCI) 314

but also the test gas used (expected to affect primarily LCI and Sacin). Given the slow 315

and linear increase in all three MBW indices in this age range, any local set of reference 316

values can be obtained by performing a reasonable number of MBW tests in normal 317

subjects at either end of the 25-65 year age range, followed by linear interpolation. 318

319

All things considered, the variability explained by age was low (Table 2), ranging 320

only 7-16% for Sacin and Scond. Despite such weak age-relationships, we did observe 321

virtually the same age-dependence of Sacin and Scond on a totally different set of 60 322

normal subjects collected in a different laboratory. Previous studies have shown Sacin 323

and Scond to be reproducible (e.g.,[15]), and the inter-subject differences in Sacin and 324

Scond that are not related to age should be interpreted in terms of inter-subject 325

differences in lung structure. Based on the fundamental principles laid out in the online 326

supplement, it is reasonable to assume that Sacin and Scond are particularly 327

associated with subject-specific heterogeneity of lung structures in a way that is very 328

different from spirometric volumes. For instance, Sacin is critically dependent on 329

average acinar branching asymmetry as well as on parallel variability in branching 330

asymmetry,[33] which is likely to be different between subjects despite similar overall 331

lung architecture. The same is probably true for Scond, which is subject to 332

heterogeneity in pressure-volume characteristics of lung units larger than acini and 333

heterogeneity in bronchomotor tone of conductive airways subtending these.[9] 334

335

Two computational sources of variability of phase III slope analysis, particularly in 336

normal subjects are (1) relatively flat phaseIII slopes, and (2) perturbations from 337

cardiogenic oscillations. These phaseIII features are particularly challenging when 338

trying to automate computation, but we show here that this is certainly feasible (Figure 339

14

4). The fact that in lung disease phaseIII slopes are greater and cardiogenic oscillations 340

are mostly absent, makes automated slope computation a realistic option with a 341

perspective of practical clinical applicability. In the early days of computing power, 342

automation of SBW phaseIII slope calculation was pursued with the perspective of using 343

it as an epidemiological tool.[28] In the SBW, automation was aimed at determining the 344

break point between phaseIII and phaseIV,[28] while for the MBW the issue is finding 345

the break point between phaseII and phaseIII.[29] 346

347

In summary, it has long been known that non-invasive measurement of gas 348

concentrations at the mouth of a patient can provide important information about 349

ventilation heterogeneity, and be a marker of structural abnormality in the small airways. 350

A consistent finding in the present MBW study is that ventilation becomes gradually 351

more heterogeneous as adults grow older between the age of 25 and 65 years, both in 352

conductive and acinar lung zones. The extent of age-dependence with respect to 353

intrinsic variability of the MBW indices, clearly shows that age is a crucial parameter to 354

be factored in when using these MBW indices as diagnostic tools. Finally, from the 355

methodological issues covered in the online supplement we can summarize that the re-356

inspired instrumental dead space volume plays a predictable role, with the most marked 357

effect on LCI, and that the MBW analysis in terms of Sacin and Scond - requiring 358

phaseIII slope determination - can be automated. 359

360

361

362

15

363

ACKNOWLEDGEMENTS 364 365

This study was funded by the Fund for Scientific Research – Flanders and supported by 366

the National Institute of Health Research (NIHR) Respiratory Disease Biomedical 367

Research Unit at the Royal Brompton and Harefield NHS Foundation Trust and Imperial 368

College London. Dr O S Usmani is a recipient of an NIHR Career Development 369

Fellowship. B.R. Thompson also received financial support through the National Health 370

and Medical Research Council of Australia (grant # 486101 & 491103). 371

372

16

373

REFERENCES 374

375

1. Burgel PR. The role of small airways in obstructive airway diseases. Eur Respir 376

Rev 2011;20:23-33. 377

2. Usmani OS, Barnes PJ. Assessing and treating small airways disease in asthma 378

and chronic obstructive pulmonary disease. Ann Med 2011 Jun 17. [Epub ahead of 379

print] 380

3. Aurora P. Multiple-breath inert gas washout test and early cystic fibrosis lung 381

disease.Thorax 2010;65:373-4. 382

4. Macleod KA, Horsley AR, Bell NJ, et al. Ventilation heterogeneity in children with 383

well controlled asthma with normal spirometry indicates residual airways disease. 384

Thorax 2009;64:33-7. 385

5. Horsley AR, Gustafsson PM, Macleod KA, et al. Lung clearance index is a 386

sensitive, repeatable and practical measure of airways disease in adults with cystic 387

fibrosis. Thorax 2008;63:135-40. 388

6. Becklake MR. A new index of the intrapulmonary mixture of inspired air. Thorax 389

1952;7:111-6. 390

7. Aurora P, Kozlowska W, Stocks J. Gas mixing efficiency from birth to adulthood 391

measured by multiple-breath washout. Respir Physiol Neurobiol 2005;148:125-39. 392

8. Crawford AB, Makowska M, Paiva M, et al. Convection- and diffusion-dependent 393

ventilation maldistribution in normal subjects. J Appl Physiol 1985;59:838-46. 394

9. Verbanck S, Paiva M. Gas mixing in the airways and air spaces. Compr Physiol 395

1:809-34,2011. 396

17

10. Verbanck S, Schuermans D, Meysman M, et al. Noninvasive assessment of 397

airway alterations in smokers: the small airways revisited. Am J Respir Crit Care Med 398

2004;170:414-9. 399

11. Verbanck S, Schuermans D, Van Muylem A, et al. Conductive and acinar lung-400

zone contributions to ventilation inhomogeneity in COPD. Am J Respir Crit Care Med 401

1998;157:1573-7. 402

12. Timmins S, King GG, Schoeffel R, et al. The relationship between emphysema 403

extent and ventilation heterogeneity. Am J Respir Crit Care Med 2011;183:A5786. 404

13. Verbanck S, Schuermans D, Paiva M, et al. The functional benefit of anti-405

inflammatory aerosols in the lung periphery. J Allergy Clin Immunol 2006;118:340-6. 406

14. Verbanck S, Schuermans D, Vincken W. Inflammation and airway function in the 407

lung periphery of patients with stable asthma. J Allergy Clin Immunol 2010;125:611-6. 408

15. Downie SR, Salome CM, Verbanck S, et al. Ventilation heterogeneity is a major 409

determinant of airway hyperresponsiveness in asthma, independent of airway 410

inflammation. Thorax 2007;62:684-9. 411

16. Sonnappa S, Bastardo CM, Wade A, et al. Symptom-pattern phenotype and 412

pulmonary function in preschool wheezers. J Allergy Clin Immunol 2010;126:519-26. 413

17. Sandqvist L, Kjellmer I. Normal values for the single breath nitrogen elimination 414

test in different age groups. Scand J Clin Lab Invest 1960;12:131-5. 415

18. Stănescu D, Teculescu D, Păcuraru R. Reproducibility and normal values of the 416

single breath nitrogen test. Scand J Respir Dis 1968;49:322-30. 417

19. Guy HJ, Prisk GK, Elliott AR, Deutschman RA 3rd, West JB. Inhomogeneity of 418

pulmonary ventilation during sustained microgravity as determined by single-breath 419

washouts. J Appl Physiol 1994;76:1719-29. 420

18

20. Buist AS, Ghezzo H, Anthonisen NR, et al. Relationship between the single-421

breath N test and age, sex, and smoking habit in three North American cities. Am Rev 422

Respir Dis 1979;120:305-18. 423

21. Knudson RJ, Lebowitz MD, Burton AP, et al. The closing volume test: evaluation 424

of nitrogen and bolus methods in a random population. Am Rev Respir Dis 425

1977;115:423-34. 426

22. Sixt R, Bake B, Oxhöj H. The single-breath N2-test and spirometry in healthy 427

non-smoking males. Eur J Respir Dis 1984;65:296-304. 428

23. Roberts CM, MacRae KD, Winning AJ, Adams L, Seed WA. Reference values 429

and prediction equations for normal lung function in a non-smoking white urban 430

population. Thorax 1991;46:643-50. 431

24. Gillooly M, Lamb D. Airspace size in lungs of lifelong non-smokers: effect of age 432

and sex. Thorax 1993;48:39-43. 433

25. Waters B, Owers-Bradley J, Silverman M. Acinar structure in symptom-free 434

adults by Helium-3 magnetic resonance. Am J Respir Crit Care Med 2006;173:847-51. 435

26. Horsley AR, Macleod KA, Robson AG, Lenney J, Bell NJ, Cunningham S, 436

Greening AP, Gustafsson PM, Innes JA. Effects of cystic fibrosis lung disease on gas 437

mixing indices derived from alveolar slope analysis. Respir Physiol Neurobiol 438

2008;162:197-203. 439

27. Verbanck S, Paiva M, Schuermans D, et al. Relationships between the lung 440

clearance index and conductive and acinar ventilation heterogeneity. J Appl Physiol. 441

2012;112:782-90. 442

28. Craven N, Sidwall G, West P, McCarthy DS, Cherniack RM. Computer analysis 443

of the single-breath nitrogen washout curve. Am Rev Respir Dis 1976;113:445-9. 444

29. Stuart-Andrews C, Kelly C, Sands S, et al. Automated detection of the phase III 445

slope during inert gas washout testing.. J Appl Physiol 2012;112:1073-81. 446

19

30. Emery MJ, Hildebrandt J, Hlastala MP. Ventilation heterogeneity in excised 447

lobes: effect of tidal volume. J Appl Physiol 2000;88:1659-71. 448

31. Janssens JP, Pache JC, Nicod LP. Physiological changes in respiratory function 449

associated with ageing. Eur Respir J 1999;13:197-205. 450

32. Bouhuys A. Pulmonary nitrogen clearance in relation to age in healthy males. J 451

Appl Physiol 1963;18:297-300. 452

33. Dutrieue B, Vanholsbeeck F, Verbanck S, et al. A human acinar structure for a 453

realistic simulations of the alveolar slope. J Appl Physiol 2000;89:1859-67. 454

20

Table 1 : Group data

Core Data set (n=120)

Supplemental Data set (n=60)

Female (n=60)

Male (n=60)

Female(n=30)

Male (n=30)

Avg SD (°) Avg SD (°) avg SD avg SD Age (years) 45.1 11.3 44.5 11.3 44.8 12.5 42.4 10.1 Height (cm) 166 5 179 7 162 7 172 18 Weight (kg) 63 9 80 11 65 16 78 17 FEV1 (L) 2.9 0.4 4.1 0.8 2.7 0.6 3.8 0.6 FVC (L) 3.6 0.5 5.2 0.9 3.4 0.6 5.0 0.8

FEV1/FVC (%) 79% 5% 79% 4% 79% 7% 76% 6% MBW

FRC (ml) 2839 491 3800 807 2673 585 3585 873

LCI (-) 6.26 0.44 6.28 0.39 5.77 0.50 5.65 0.49

Scond (L-1) 0.035 0.014 0.034 0.010 0.036 0.018 0.036 0.015

Sacin (L-1) 0.083 0.030 0.099 0.032 0.086 0.037 0.091 0.041 SBW

SnIII_FRC (L-1) 0.059 (0.052 - 0.069) 0.050 (0.043 - 0.058)

SnIII_RV (L-1) 0.053 0.020 0.046 0.016

SIII_RV (%/L) 1.16 (1.04 - 1.47) 1.04 (0.89 - 1.12) FEV1: forced expiratory volume in one second; FVC: forced vital capacity; LCI : lung clearance index, Scond ,Sacin; MBW index of conductive and acinar ventilation heterogeneity. SnIII_FRC,SnIII_RV : normalized phaseIII slope of a SBW maneuver starting from functional residual capacity or residual volume; SIII_RV : unnormalized phase III slope of a SBW maneuver starting from RV.

(°) when a variable was not normally distributed, median (95% confidence interval) is shown instead.

21

Table 2

Regression equations for MBW indices from the core data set (n=120).

Adjusted

R2

ResidualStandardDeviation

MBW

LCI (-) M/F pooled 0.0223 * age + 5.275 0.36 0.330

Scond (L-1) M/F pooled 0.000358 * age + 0.0187 0.10 0.0116

Sacin (L-1) F 0.00078 * age + 0.0482 0.07 0.0291

M 0.00118 * age + 0.0472 0.16 0.0294 LCI : lung clearance index, Scond ,Sacin; MBW index of conductive and acinar ventilation heterogeneity.

22

Table 3

Regression equations for SBW indices from the core data set (n=120).

Adjusted

R2

ResidualStandardDeviation

SnIII_FRC (L-1) F e (0.0227*age-0.583*FRC+2.08*Height-5.955)

0.60 0.297

M e (0.0363*age-0.537*FRC -2.644)

0.72 0.357 SnIII_RV

(L-1)

M/F

0.000646*age + 0.00534 * FRC + 0.0372

0.18

0.016

SIII_RV (%/L) F e (0.0204*age - 0.726)

0.29 0.342

M e (0.0285*age - 1.325) 0.48 0.307 Age in years, Height in metres, and FRC is functional residual capacity in Litres; SnIII_FRC,SnIII_RV : normalized phase III slope of a SBW maneuver starting from FRC or RV; SIII_RV : unnormalized phase III slope of a SBW maneuver starting from RV.

23

FIGURE LEGENDS

Figure 1 : Scatterplots of Sacin versus age.

Panel A: Core data sets with corresponding regression lines for females (open symbols;

dashed line) and males (closed symbols; solid line).

Panel B: Core data set pooling females and males (closed squares; dotted line) and

supplementary data set (crosses; solid line) with corresponding regression lines.

Figure 2 : Scatterplots of Scond versus age. Same representation as in Figure 1.

24

Figure 3 : Scatterplots of LCI versus age. Same representation as in Figure 1.

Figure 4 : Bland-Altman plots comparing Sacin, Scond, FRC and LCI values computed

either manually or with an automated procedure.