Supplementary Online Material A randomised study of augmentation therapy in alpha-1 antitrypsin deficiency: exploring the role of CT densitometry A. Dirksen, E. Piitulainen, D.G. Parr, C. Deng, M. Wencker, S.B. Shaker and R.A. Stockley METHODS Inclusion/exclusion criteria Eligible patients were at least 18 yrs of age, had a history of at least one exacerbation in the past 2 yrs, and had a post-bronchodilator forced expiratory volume in 1 s (FEV1) ≥25% and ≤80% predicted and a ratio of post-bronchodilator FEV1 to slow vital capacity (VC) ≤0.70. Patients with normal spirometry could be included if a carbon monoxide transfer coefficient (KCO) was ≤80% of the predicted value. Patients were excluded from the study if they had a post-bronchodilator FEV1 <25% predicted or body weight <42 kg or >92 kg, had had any lung surgery or augmentation therapy within the past 2 yrs, had a history of lung transplantation or were on the waiting list for any thoracic surgery, and had smoked during the past 6 months or were plasma-positive for cotinine. Randomisation and blinding When all appropriate study entrance criteria had been met, subjects were randomised in a 1:1 ratio to receive Prolastin® or placebo. A computer-generated random code was 1 used to produce randomisation envelopes that were issued to the unblinded pharmacist or designee at each study centre and which were to be kept confidential. The randomisation envelopes were sent to the pharmacist with the study medication. The randomisation numbers were assigned to subjects in ascending order at the baseline visit, when the subject’s eligibility had been confirmed. Several measures were taken to ensure blinding, both with regard to the study drug and to the assessment of efficacy results. Blinding of different study groups was guaranteed by ensuring that all subjects received the same total volume per kg body weight of study medication with no visible difference in the external aspect between Prolastin® and placebo (variation in colour by lot was masked by using opaque sleeves). Throughout the course of the study, individual treatment assignments were unknown to the treating investigators and nurses, the subjects, the clinical management and monitoring team, the central computed tomography (CT) scan facility, and the sponsor’s data management, clinical and biostatistical teams. Every effort was made to maintain the integrity of the blinding through locking of the database. The randomisation block size was not disclosed to the study sites. During the course of the study, the randomisation code was maintained in a secure fashion and was made available only to designated unblinded team members, which included the clinical site pharmacy personnel who prepared the study medication and the BeroSearch monitor responsible for monitoring the pharmacy. 2 CT densitometry CT image acquisition In all three centres a low-dose volumetric scanning protocol was used. Multidetector array CT scans of the chest were performed in the supine position and during approximately 10 s breath hold. Patients were told to inspire as close to total lung capacity as possible in order to improve reproducibility and reduce low-density artefacts arising from air-trapping. Scanning was performed 30 min (and within a maximum of 4 h) after inhaled bronchodilator therapy. Three deep inspiratory manoeuvres were performed immediately prior to scanning to ensure standardised expansion and ventilation of basal areas that are susceptible to atelectatic changes. All scanning was performed in a caudo-cranial direction in order to reduce artefacts arising from diaphragm movement, and with the arms raised above the head to reduce densitometric artefact arising from X-ray spectral changes and beam scatter due to juxtaposition of dense tissue adjacent to the lungs. No contrast medium was injected. All raw data were reconstructed using an edge-smoothing image reconstruction algorithm and were saved in DICOM format at the time of imaging. Data were transferred to CD for shipment to the central facility for densitometric analysis. Scan acquisition parameters were standardised, taking account of the differences in scanner manufacturers and models that existed between the three centres. The minor differences in scanner settings are shown in table 1. The scanner model used at each site was consistent throughout the study. Preferred scanning parameters were 140 kVp, 40 mA and pitch 1.5, with reconstructed slice thickness of 5 mm and with an increment of 2.5 mm. Radiation per CT scan was low at around 1 mSv. For the three different scanners, detailed protocol instructions were provided by MEDIS Medical Imaging Systems BV, Leiden, The Netherlands, who also verified adherence to the CT 3 acquisition guidelines. Quality assurance and the independent, blinded quantitative analysis of CT images were undertaken by a central facility (Heart Core Global Medical Imaging). Scanner calibration Mandatory scanner air calibration was performed according to the scanner manufacturers’ instructions within 3 h of the first patient scan and every 3 h during scanning lists. Mandatory water calibration was performed by the manufacturers (using the manufacturers’ water phantom) at least every 3 months using the clinical scan protocol. These two processes are intended to ensure that scanner Hounsfield numbers for air and water are measured at -1000 Hounsfield units (HU) and 0 HU for different scan protocols, including the study protocol. Lung density values lie between these two points and, if linearity is assumed, adjustment of air and water numbers should ensure validity of lung values. Phantom analysis A dedicated Perspex and foam phantom was used for additional quality assurance data. Phantom scans were performed prior to site initiation and prior to the first patient scan at each site, and every 6 months throughout the study. Additional phantom scans were performed following any scanner technical changes, e.g. X-ray tube replacement. Densitometric assessment of foam from phantom images generated numbers that were averaged to provide a single figure for evaluation of scanner calibration. 4 Site assessment Site assessment, including descriptive data relating to the scanner model, was recorded on the Core Laboratory Questionnaire and phantom scan (see above). An additional visit to the Copenhagen centre for more detailed scanner assessment was necessary prior to initiation. Scanner software at this site had an unusually high cut-off for low voxel densities at -1000 HU, which prevented the use of “main air density” for densitometric adjustment as an integral component of patient scan analysis, and a satisfactory method of acquiring appropriate internal air calibration data at this site could not be identified. This software limitation was a recognised feature of the scanner model at Copenhagen, and it was agreed before study commencement that no air internal calibration and only the blood internal calibration method [1] would be used for imaging acquired at the Copenhagen centre. Scan analysis Lung imaging was analysed centrally (Heart Core Global Medical Imaging) using semiautomated software (Pulmo-CMS, MEDIS Medical Imaging Systems BV, Leiden, The Netherlands) as described previously [2]. This program delineates lung parenchyma by automatic lung segmentation (defined by an upper threshold of -380 HU) using a method known as the “seeded-region growing technique”. The trachea is located manually, and the program automatically identifies low-density voxels below the chosen threshold value that are contiguous with the tracheal origin. Exclusion of the trachea between the point of seeding and the carina then occurs automatically after lung segmentation. Internal calibration allows adjustment of densitometric values by re-scaling lung segmentation according to measurements obtained from the image series. Measurement of blood density in the subdiaphragmatic descending thoracic aorta [1] and air density in the pre- 5 ventral area [3] were used for internal calibration on image sequences acquired at the Birmingham and Malmö centres. Measurement of blood density in the subdiaphragmatic descending thoracic aorta was used for internal calibration at the Copenhagen centre (see above). Blood calibration is performed semi-automatically commencing with the manual location of markers within the aortic lumen to define the upper and lower limits of sampling. A circular region of interest of approximately 5000 voxels is then automatically located within the aortic lumen on each image between the defined upper and lower boundaries, and the mean blood density (in HU) is calculated [4]. The same process is applied to the pre-ventral air for internal air calibration [3]. Automatic detection of the septum delineates right and left lungs, allowing calculation of parameters for both lungs individually and as a whole. Errors in contour detection, such as inclusion of oesophagus or bowel, or errors in left/right labelling require manual correction using a computer “mouse” as scribe. There is an inherent potential for error secondary to variability in operator performance [1], although this error is small and analysis at a central facility (Heart Core) would be expected to minimise variability. Blinded review of scan data All CT scan data were assessed prior to study analysis by a blinded review panel of experts (D. Parr, M. Brantly, A. Dirksen, T. Grobben). The panel carried out visual inspection of images to identify any densitometric values that were not reliable because of technical issues such as scan acquisition and image quality, and confounding factors such as recent exacerbations, pneumonia, heart failure and presence of bullous lung. Decisions were made by mutual consent and were fully documented. Any CT scan that was considered invalid was excluded from the primary efficacy analysis. Reasons for 6 exclusion of scans were: (a) patient discontinued from the study prior to month 12 such that no post-baseline scan was available; (b) a problem with the scanning procedure, the scanner or the calibration of the scanner was detected; and (c) identification of a pulmonary process unrelated to emphysema progression that affected lung density. Parameter for CT end-points Based on previous studies [5–7] and recommendations of an expert review [8], the 15th percentile point was chosen as the parameter for the CT end-points and expressed as the 15th percentile density (PD15). The 15th percentile point is defined as the value (HU) at which 15% of the voxels in the frequency distribution histogram have a lower density [6, 9], and it may be expressed as PD15 (g/L) by the simple addition of 1000 to the Hounsfield value of the 15th percentile point. Calculation of TLC-adjusted lung density TLC-adjusted 15th percentile of lung density was calculated as follows: Adjusted LD = Observed LD x (Observed TLV / Predicted TLC) where “LD” represents 15th percentile of lung density measured by CT scan and “TLV” represents lung volume achieved with full inspiratory effort during scan acquisition (measured from whole CT imaging series). Predicted TLC is estimated from: 7.99 x (height in m) 7.08 for males, and 6.60 x (height in m) 5.79 for females [10]. 7 RESULTS CT scan data at 3 months Optional additional scans were performed at 3 months with the intention of identifying any initial influence of treatment. No difference was found between the Prolastin® and placebo groups over this short time period. Overall mean (SD) TLC-adjusted PD15 values for changes from baseline at 3 months were 1.37 g/L (4.09) in the Prolastin® group and -0.29 g/L (2.77) in the placebo group (p=0.159). Three-month data were only available from 36 of 71 participants in the modified intent-to-treat (mITT) population and from two of the three centres. These data were not included in the final analysis of any treatment effect. Change in lung volume and lung weight Mean values of lung volume as assessed by CT remained almost unchanged over the course of the study (table 2). Comparison of the slopes showed that there was a greater decrease in lung weight in the placebo group, although the treatment difference was not statistically significant (table 3). Change in lung function end-points Mean annual decreases in FEV1 were -0.043 L (95% CI -0.063– -0.024) and -0.023 L (0.043– -0.004) in the Prolastin® and placebo groups, respectively (p=0.147); for diffusing capacity of lung for carbon monoxide (DLCO), the corresponding annual decreases were -0.460 mmol/min/kPa (95% CI -0.603– -0.317) and -0.343 mmol/min/kPa (-0.489– -0.196) (p=0.257). There was a large variation in changes in KCO among individual patients. The mean annual changes in KCO were nearly identical for the Prolastin® and placebo groups 8 (-0.036 mmol/min/kPa/L (95% CI -0.051– -0.020) and -0.035 mmol/min/kPa/L (-0.051– 0.020), respectively; p=0.967). Serious adverse events (SAEs) A listing of the numbers of subjects with at least one SAE is shown in table 4. DISCUSSION Additional discussion of the physiological adjustment method to correct lung density for variation in inspiratory level In Methods 1 and 3, adjustment of lung densities is physiologically based on the assumption that the lung behaves like a sponge (sponge-model) [11]. The “sponge model” makes two assumptions: weight = volume × density, and the total lung weight stays constant during the breathing cycle, which means that, for example, halving the volume should double the density. The latter assumption may seem incompatible with the fact that the generation of negative intrathoracic pressure required for inspiration will necessarily be accompanied by a variable increase in blood flow to the lung. However, most of the blood flow into the lung is intercepted by large capacitance veins and, with a threshold for the soft tissue–lung interface of -380 HU, large vessels were excluded from the “lung” by the region-growing algorithm. Consequently, this physiological effect should not influence the weight of the lung as calculated from the CT scans using the PulmoCMS software [11]. 9 REFERENCES 1 Stoel BC, Vrooman HA, Stolk J, Reiber JH. Sources of error in lung densitometry with CT. Invest Radiol 1999; 34: 303–309. 2 Stolk J, Ng WH, Bakker ME, Reiber JH, Rabe KF, Putter H, Stoel BC. Correlation between annual change in health status and computer tomography derived lung density in subjects with alpha1-antitrypsin deficiency. Thorax 2003; 58: 1027–1030. 3 Parr DG, Stoel BC, Stolk J, Nightingale PG, Stockley RA. Influence of calibration on densitometric studies of emphysema progression using computed tomography. Am J Respir Crit Care Med 2004; 170: 883–890. 4 Stoel BC, Stolk J. Optimization and standardization of lung densitometry in the assessment of pulmonary emphysema. Invest Radiol 2004; 39(11): 681–688. 5 Dirksen A, Friis M, Olesen KP, Skovgaard LT, Sorensen K. Progress of emphysema in severe alpha 1-antitrypsin deficiency as assessed by annual CT. Acta Radiol 1997; 38: 826–832. 6 Parr DG, Stoel BC, Stolk J, Stockley RA. Validation of computed tomographic lung densitometry for monitoring emphysema in alpha1-antitrypsin deficiency. Thorax 2006; 61: 485–490. 7 Parr DG, Sevenoaks M, Deng C, Stoel BC, Stockley RA. Detection of emphysema progression in alpha 1-antitrypsin deficiency using CT densitometry; methodological advances. Respir Res 2008; 9(1): 21. 8 Newell JD, Hogg JC, Snider GL. Report of a workshop: quantitative computed tomography scanning in longitudinal studies of emphysema. Eur Respir J 2004; 23: 769– 775. 9 Dirksen A, Dijkman JH, Madsen F, Stoel B, Hutchinson DC, Ulrik CS, Skovgaard LT, Kok-Jensen A, Rudolphus A, Seersholm N, Vrooman HA, Reiber JH, Hansen NC, 10 Heckscher T, Viskum K, Stolk J. A randomized clinical trial of alpha 1-antitrypsin augmentation therapy. Am J Respir Crit Care Med 1999; 160: 1468–1472. 10 Quanjer PH, Tammeling GJ, Cotes JE, Pedersen OF, Peslin R, Yernault JC. Lung volumes and forced ventilatory flows. Report Working Party Standardization of Lung Function Tests, European Community for Steel and Coal. Official Statement of the European Respiratory Society. Eur Respir J Suppl 1993; 16: 5–40. 11 Shaker SB, Dirksen A, Laursen LC, Skovgaard LT, Holstein-Rathlou NH. Volume adjustment of lung density by computed tomography scans in patients with emphysema. Acta Radiol 2004; 45: 417–423. 11 TABLE 1 Scanner settings at the three active study centres Parameter Birmingham Copenhagen Malmö GE Philips Siemens Lightspeed MX 8000 jdt Somotom sensation 4 16 16 kVp 140 140 140 mA 40 60 60 mAs 20 eff.mAs=20/ mAs=30# eff.mAs=20/ mAs=30# 0.5 0.5 0.5 6 24 24 Beam pitch mm 1.5 1.5 1.5 Collimation mm 4x5 16 x 1.5 16 x 1.5 5 5 5 Increment mm 2.5 2.5 2.5 Filter Soft A B10f CT scanner manufacturer Type Detector rows n Image acquisition Rotation time s Detector pitch mm Slice thickness mm CT: computed tomography. #: scanners Copenhagen and Malmö were using the so-called effective mAs. Instead of effective mAs = 20 the actual mAs was 30. 12 TABLE 2 Changes (in litres) in CT-measured total lung volume from baseline (mITT population) Prolastin® Placebo (n=36) (n=35) Month 12 (n=34, 33) -0.029±0.390 -0.104±0.414 Month 24 (n=35, 32) -0.011±0.376 0.041±0.277 Month 30 (n=18, 16) -0.005±0.356 -0.045±0.284 0.002 (-0.049–0.053) 0.004 (-0.049–0.057) Statistic Change from baseline mean±SD Random coefficient model# mean slope¶ 95% CI Estimated treatment difference in mean slopes 95% CI p-value for treatment difference+ -0.002 (-0.075–0.071) 0.959 CI: confidence interval. #: lung volume (L) as dependent variable, treatment, centre, treatment-by-time interaction as the fixed effects, and intercept and time as the random effects; ¶: average annual loss of lung volume (L); +: Prolastin® minus placebo. 13 TABLE 3 Changes in CT-measured lung weight (g) from baseline (mITT population) Prolastin® Placebo (n=36) (n=35) Month 12 (n=34, 33) -6.028±31.283 -17.963±31.853 Month 24 (n=35, 32) -8.631±44.524 -18.426±27.636 Month 30 (n=18, 16) -13.210±45.342 -14.644±32.441 -4.901 (-10.196–0.394) -9.770 (-15.275– -4.266) Statistic Change from baseline mean± SD Random coefficient model# mean slope¶ 95% CI Estimated treatment difference in mean slopes 95% CI p-value for treatment difference+ 4.869 (-2.768–12.507) 0.207 CI: confidence interval. #: lung weight (g) as dependent variable, treatment, centre, treatment-by-time interaction as the fixed effects, and intercept and time as the random effects; ¶: average annual loss of lung weight (g); +: Prolastin® minus placebo. 14 TABLE 4 Numbers of subjects with at least one SAE Prolastin® Placebo n=38 n=39 n (%) n (%) 28 40 10 (26) 18 (46) Severe exacerbations# 5 (13) 6 (15) Pneumonia 3 (8) 4 (10) Pneumothorax 2 (5) 0 Atrial fibrillation 2 (5) 0 Biliary colic 1 (3) 0 Constipation 1 (3) 0 Epistaxis 1 (3) 0 Gall bladder disorder 1 (3) 1 (3) Gastro-oesophageal reflux 1 (3) 0 Malaria 1 (3) 0 Menorrhagia 1 (3) 0 Psoriasis 1 (3) 0 Transient ischaemic attack 1 (3) 0 Upper limb fracture 1 (3) 0 Abdominal pain 0 1 (3) Intra-abdominal haemorrhage 0 1 (3) Rectal haemorrhage 0 1 (3) Nodule 0 1 (3) Cholestatic jaundice 0 1 (3) Appendicitis 0 1 (3) Sepsis 0 1 (3) Adverse event Total number of SAEs Number of patients with any SAE 16 Subcutaneous abscess 0 1 (3) Urinary tract infection 0 1 (3) Arthralgia 0 1 (3) Osteoarthritis 0 1 (3) Breast cancer 0 1 (3) Chronic obstructive pulmonary 0 1 (3) Dyspnoea 0 1 (3) Pleural effusion 0 1 (3) Pulmonary embolism 0 2 (5) Pulmonary oedema 0 1 (3) Lichen sclerosus 0 1 (3) disease SAE: serious adverse event. #: severe exacerbation with hospitalisation not previously reported as SAE. 17