laboratory and animal investigations
`Comparison of Breath-Enhanced to
`Breath-Actuated Nebulizers for Rate,
`Consistency, and Efficiency*
`
`Kitty Leung, BSc; Emily Louca, BSc, RRT; and
`Allan L. Coates, B Eng(Elect), MDCM
`
`Objectives: To evaluate differences between three new-generation nebulizers—Pari LC Star
`(Pari Respiratory Equipment; Mississauga, ON, Canada), AeroEclipse (Trudell Medical Interna-
`tional, London, ON, Canada), and Halolite (Medic-Aid Limited, West Sussex, UK)—in terms of
`rate and amount of expected deposition as well as the consistency of the doses delivered.
`Methods: The in vitro performance characteristics were determined and then coupled to the
`respiratory pattern of seven patients with cystic fibrosis (age range, 4 to 18 years) in order to
`calculate expected deposition. The Pari LC Star and AeroEclipse were characterized while being
`driven by the Pari ProNeb Ultra compressor (Pari Respiratory Equipment) for home use, and by
`a 50-psi medical air hospital source. The Halolite has its own self-contained compressor.
`Algorithms for the rate of output for the inspiratory flow were developed for each device. Patient
`flow patterns were divided into 5-ms epochs, and the expected deposition for each epoch was
`calculated from the algorithms. Summed over a breath, this allowed the calculation of the
`estimated deposition for each patient’s particular pattern of breathing.
`Results: The rate of deposition was highest for the Pari LC Star and lowest for the Halolite. Rate
`of deposition was independent of respiratory pattern for the Pari LC Star and AeroEclipse, but
`proportional to respiratory rate for the Halolite. The differences between the Pari LC Star and
`AeroEclipse were less when driven by the 50-psi source. The AeroEclipse had the least amount
`of drug wastage. As designed, the Halolite delivered a predetermined amount of drug very
`accurately, whereas expected deposition when run to dryness of the other two devices had
`significant variations.
`Conclusions: To minimize treatment time, the Pari LC Star would be best. To minimize drug
`wastage, the AeroEclipse would be best. To accurately deliver a specific drug dose, the Halolite
`would be best.
`(CHEST 2004; 126:1619 –1627)
`
`Key words: aerosols; asthma; breath-actuated nebulizers; breath-enhanced nebulizers; cystic fibrosis; pediatrics
`
`Abbreviations: CF ⫽ cystic fibrosis; CI ⫽ confidence index; Ot ⫽ total drug output; RF ⫽ respirable fraction;
`UV ⫽ ultraviolet; Vr ⫽ residual volume
`
`J et nebulization is one of the mainstays of treat-
`
`ment for cystic fibrosis (CF), where it is used to
`deliver medications ranging from antibiotics1 to mu-
`
`colytics,2,3 and is also commonly used to deliver
`bronchodilators
`for
`the emergency department
`treatment of asthma.4 From previous studies,5–7
`
`*From the Division of Respiratory Medicine and Lung Biology
`Research, Hospital for Sick Children, Research Institute, To-
`ronto, ON, Canada.
`The nebulizers studied were provided through the generosity of
`PARI Respiratory Equipment Inc., Trudell Medical International
`Inc., and Medic-Aid Limited.
`Supported from a grant from the Hospital for Sick Children’s
`Foundation, made possible by a generous donation from Arnold
`and Lynn Irwin for cystic fibrosis research.
`
`Manuscript received September 5, 2003; revision accepted May
`28, 2004.
`Reproduction of this article is prohibited without written permis-
`sion from the American College of Chest Physicians (e-mail:
`permissions@chestnet.org).
`Correspondence to: Allan L. Coates, B Eng(Elect), MDCM,
`Division of Respiratory Medicine, The Hospital for Sick Children,
`555 University Ave, Toronto, ON, Canada, M5G 1X8; allan.
`coates@sickkids.ca
`
`www.chestjournal.org
`
`CHEST / 126 / 5 /NOVEMBER, 2004
`
`1619
`
`Liquidia's Exhibit 1103
`Page 1
`
`

`

`breath-enhanced nebulizers are more efficient than
`unvented nebulizers, but not all breath-enhanced
`nebulizers have the same efficiency,8 with differ-
`ences in residual volume (Vr) and particle size
`resulting in significant differences in expected pul-
`monary deposition. There is a new generation of jet
`nebulizers that are breath actuated, producing med-
`ication only during inspiration, which makes them
`potentially even more efficient than the breath-
`enhanced devices. At present, there are little com-
`parative data available to help the clinician choose
`between devices for specific applications.
`All jet nebulizers have a nebulizing chamber con-
`taining liquid medication and a high-pressure, high-
`velocity jet of gas that creates a partial vacuum at the
`exit orifice of the jet, resulting in the medication
`being drawn up toward the high-velocity orifice,
`where shear forces fragment the liquid into a poly-
`disperse aerosol. The aerosol passes around a series
`of baffles, where larger particles are removed by
`inertial impaction and fall back into the reservoir for
`renebulization. Particles that escape the baffles ei-
`ther leave the nebulizer, or “rain out” and fall back
`into the medication chamber under the influence of
`gravity. Simplistically, the major difference between
`unvented and breath-enhanced nebulizers is that the
`patient’s inspiratory flow is entrained into the device,
`and particles that would otherwise rain out are swept
`along into the patient during inspiration.5,9 Hence,
`the rate of output of breath-enhanced nebulizers
`increases with increasing inspiratory flow and falls
`back to baseline during expiration when no flow is
`entrained. Furthermore, since inertial impaction of
`droplets on the baffles is in part dependent on
`velocity of the particle, increases in entrained flow
`increases the likelihood that larger particles will
`impact on the baffles. This may give rise to a smaller
`particle size distribution during inspiration as the
`inspiratory flow increases.8 Particles between 1 ␮m
`and 5 ␮m in diameter are ideal for pulmonary drug
`delivery, in that they are small enough so as not to be
`removed by inertial impaction at the posterior phar-
`ynx, but large enough to carry a significant amount of
`drug. Given that particle volume is proportional to
`the third power of the radius, particles ⬍ 1 ␮m carry
`little drug. The fraction of the volume of the nebu-
`lizer output carried in particles with a diameter ⱕ 5
`␮m is defined as the respirable fraction (RF).10 –13
`In terms of the appropriate choice of device, a
`number of factors come into play. Clearly, the ability
`to produce a high-density aerosol with a large RF
`during the inspiratory phase is the basic principle,
`but other factors such as Vr at end nebulization are
`an issue, especially if the medication is very expen-
`sive.14 Since one of the challenges in the treatment
`of CF is patient adherence to recommended treat-
`
`Figure 1. A schematic of the breath-enhanced nebulizer Pari
`LC Star.
`
`ment regimens, devices that reduce treatment time
`would be expected to offer advantages to the already
`very time-consuming daily multifaceted treatment
`activities of these patients.15–17 The devices should
`therefore be evaluated on the expected pulmonary
`deposition of a specific dose, and the delivery time
`required. The breath-enhanced nebulizer, the Pari
`LC Star (Pari Respiratory Equipment; Mississauga,
`ON, Canada) [Fig 1], has been shown to be one of
`the more efficient breath-enhanced nebulizers.8
`Breath-actuated nebulizers, such as the AeroEclipse
`(Trudell Medical International, London, ON, Can-
`ada) [Fig 2] and Halolite (Medic-Aid Limited, West
`Sussex, UK) [Fig 3] have recently been developed.
`The Halolite uses an adaptive aerosol delivery system
`that can adapt the drug delivery to each patient’s
`breathing pattern. Table 1 provides a functional
`comparison of all three devices.
`The purpose of this study was to compare the
`three devices in terms of in vitro performance,
`expected in vivo rate of deposition, and in vivo
`efficiency using the respiratory pattern of patients
`with CF breathing through a nebulizer. Significant
`end points are considered to be the percentage of
`the initial dose that would be delivered to the lungs,
`the time required to deliver a “target” dose, and the
`ability to deliver a precise pulmonary dose. It is
`recognized that the importance of these variables
`depends on the expense of the drug being delivered,
`the value in terms of possible greater adherence to
`recommended therapy from rapid delivery, and the
`therapeutic safety profile of the drug in terms of
`accurately delivering a specific amount.
`
`Methods and Materials
`
`Device Operation
`
`The nebulizers and compressors used in this study were the
`Pari LC Star nebulizer driven by the Pari Proneb Ultra compres-
`
`1620
`
`Laboratory and Animal Investigations
`
`Liquidia's Exhibit 1103
`Page 2
`
`

`

`Figure 2. A schematic of the breath-enhanced, breath-actuated nebulizer (BAN) AeroEclipse.
`
`sor (Pari Respiratory Equipment); the AeroEclipse nebulizer,
`which was also driven by the Pari compressor, as no specific
`compressor was recommended; and the Halolite nebulizer with a
`built-in compressor. The Halolite is a microprocessor-controlled
`
`Figure 3. A schematic of the breath-actuated nebulizer Halolite,
`which uses adaptive aerosol delivery technology. Insert shows
`breath tracing; dark areas represent device activation.
`
`device that activates the compressor on each inspiration. Three
`examples of both the Pari LC Star and the AeroEclipse were
`studied, but only a single Halolite device was available. The test
`drug was 2.5 mg (0.5 mL) of albuterol (Ventolin Respirator
`Solution; GlaxoSmithKline; Mississauga, ON, Canada) diluted
`with 3.5 mL of saline solution. This was chosen because it lends
`itself to ultraviolet (UV) spectrophotometry for quantification of
`output.8 The Pari LC Star and the AeroEclipse were also
`evaluated using compressed dry air (hospital air, 50-psi source) at
`8 L/min, which is the same flow recommended by the manufac-
`turer for the AeroEclipse. Flow from the compressor was
`measured by a flow calibration instrument (Timeter RT200;
`Allied Health Care Products, St. Louis, MO), and the flowmeters
`on the hospital air line were calibrated to adjust for “back
`pressure,”12 so as to deliver the expected driving nebulizing flow.
`When driving either the AeroEclipse or the Pari LC Star, the
`output of the ProNeb Ultra compressor was 4.9 L/min.
`
`Particle Size Distribution and Determining Nebulizer Output
`
`Both the Pari LC Star and the AeroEclipse were characterized
`in terms of particle size distribution and rate of output during
`steady-state conditions. Briefly, the device was mounted to allow
`aerosol to pass through the laser beam of a Malvern Mastersizer
`X (Malvern Instruments; Worcestershire, UK), and particle size
`was measured using the Mie theory for transparent droplets.
`Care was taken to avoid vignetting.18 This method has been
`described in detail elsewhere.19 Measurements were made after
`2 min of nebulization, which allowed the nebulizer to attain a
`steady-state temperature,19 after which particle size distribution
`and RF were calculated. In order to mimic entrained flow, air at
`40% relative humidity was added at the point of the inspiratory
`valve in flow increments of 5 L/min up to a maximum of 35
`
`www.chestjournal.org
`
`CHEST / 126 / 5 /NOVEMBER, 2004
`
`1621
`
`Liquidia's Exhibit 1103
`Page 3
`
`

`

`Table 1—A Functional Comparison of the Pari LC Star, AeroEclipse, and Halolite
`
`Pari LC Star
`
`AeroEclipse
`
`Halolite
`
`Breath enhanced
`
`Breath enhanced
`Breath actuated
`
`Inspiratory valve allows air to entrain
`into the chamber during inspiration
`when the flow of patient is greater
`than nebulizing flow
`
`When entrained flow is ⬎ 8 L/m, a
`unique spring-loaded mechanism
`allows the actuator piston to be
`pulled down onto the jet and
`nebulization commences
`Aerosol is only produced during the
`inspiratory phase, making it
`potentially very efficient
`
`Expiratory valve on mouthpiece prevents
`exhaled gases from entering the
`nebulizer
`Treatment complete when device
`sputters
`
`Expiratory valve on mouthpiece
`prevents exhaled gases from
`entering the nebulizer
`Treatment complete when device
`sputters
`
`Breath actuated
`Uses adaptive aerosol delivery system,
`which adapts drug delivery to each
`individual patient’s breathing pattern
`Device has two operating buttons; each
`is designed to deliver the
`manufacturer’s preset volume
`Aerosolization begins when the patient
`pushes the appropriate button
`(albuterol for this study) and begins
`breathing
`Halolite analyses the first three breaths
`of the patient to determine the
`breathing pattern
`A pulse of drug is delivered every
`subsequent breath only during the
`first 50% of inspiration
`No entrainment of flow on inspiration
`Output is constant for each pulse and
`independent of the inspiratory flow
`Valves divert ventilation around
`nebulizing chamber
`
`Treatment is complete when the preset
`dose has been delivered
`
`L/min, and particle size distributions were measured in each
`situation. For the AeroEclipse, the first level of entrained flow
`was 8 L/min because the spring-loaded valve only opens when
`entrained flow reaches this level. The microprocessor control of
`the Halolite makes conventional particle sizing difficult since the
`device is not designed to run continuously. This intermittent
`operation results in differences in temperature of the aerosol
`when nebulized continuously vs pulsed. The increased accuracy
`of 2,000 sweeps during data gathering by the Malvern Master-
`sizer X for particle size distribution calculations in “continuous”
`mode offsets the limited data achieved from a “pulse,” even with
`differences in temperature of the aerosol being particle sized. To
`create a continuous mode, the device was dismantled and the
`back pressure created by the compressor when driving the
`Halolite handset, which contains the nebulizing device and
`microprocessor, was measured as 28 to 30 psi. The compressor
`uses an elastic reservoir that allows pressure to increase during
`expiration, and contributes to the compressor output during the
`pulse of aerosol. This resulted in a driving pressure that is
`considerably higher than that which would have occur if the
`compressor were driving the nebulizer continuously. The micro-
`processor within the Halolite handset was dismantled, and the
`nebulizer was driven by a dry air gas source at a flow matching
`the back pressure previously measured from the Halolite com-
`pressor, which resulted in an output flow of 5.4 L/min. The
`mouthpiece of the handset was positioned to send a continuous
`stream of aerosol across the laser beam. Since there is no
`entrained flow, only one measurement condition was necessary.
`Prior to the particle size measurements, devices were weighed
`empty (for the Halolite, this was only the medication chamber),
`filled, and reweighed using an electronic balance (BL150; Sarto-
`rius Corporation; Edgewood, NY). After 4 min of steady-state
`output, the devices were reweighed. Changes in drug concentra-
`tion due to evaporative losses were assessed initially by changes in
`UV spectrophotometry and water vapor pressure osmolarity
`
`(Advanced Micro-Osmometer 3300; Advanced Instruments;
`Norwood, MA). Eventually, only osmolarity was used since the
`simpler technique gives identical results to the more complex UV
`spectrophotometry. The drug output over the nebulization pe-
`riod was calculated from the Vr and the changes in concentra-
`tion, as seen in Appendix 1.
`For each 4-min run under each condition of entrained flow, the
`total rate of output and that in the RF was calculated, and the
`mean taken for the three examples of both the Pari LC Star and
`the AeroEclipse. Polynomial curve-fitting techniques were used
`to create the algorithm for the rate of output—total and within
`the RF— over the range of entrained flow. Finally, both the Pari
`LC Star and the AeroEclipse were run to dryness, defined as the
`absence of mist for at least 10 s,8,12 to allow the calculation of the
`total output of the device. The details of these techniques have
`been described.8,19,20 Output data for the Halolite were collected
`by connecting it to a modified Harvard pump (Model 613;
`Harvard Apparatus; Holliston, MA) that delivered two half-sine
`waves, with an inspiratory time/total time of respiratory cycle
`(Ti/Ttot) of 0.4, a tidal volume of 500 mL, and a respiratory rate
`of 20 breaths/min. These settings approximate the tidal volumes
`and timing of actual patient flow traces (see below). The Harvard
`pump was run until the Halolite sensed that the preset volume
`had been delivered. The output was calculated from the drug
`remaining in the nebulizer cup via gravimetric techniques and
`changes in osmolarity, which had complete agreement with UV
`spectrophotometry. When the output multiplied by the RF is
`divided by the number of breaths, the result is the expected
`deposition per breath.
`
`Calculation of Estimated Pulmonary Deposition
`
`From a previous study,21 digitized breath tracings of seven
`patients with CF (age range, 4 to 18 years) breathing through a
`nebulizer (Table 2) were used. Patients with FEV1 values ⬎ 60%
`
`1622
`
`Laboratory and Animal Investigations
`
`Liquidia's Exhibit 1103
`Page 4
`
`

`

`Table 2—Demographic Information on the Seven
`Patients Whose Breathing Patterns Were Used To
`Calculate Estimated Pulmonary Deposition and In
`Vivo Efficiency*
`
`Patient
`No.
`
`Age,
`yr
`
`Height,
`cm
`
`Weight,
`kg
`
`FEV1, %
`Predicted
`
`Respiratory Rate,
`Breaths/min
`
`138
`11
`1
`141
`11
`2
`122
`7
`3
`174
`18
`4
`122
`7
`5
`142
`14
`6
`159
`4
`7
`*From Coates et al.21
`
`28
`34
`22
`55
`23
`31
`42
`
`84
`76
`68
`113
`78
`26
`55
`
`17.1
`18.7
`31.4
`18.9
`41.6
`40.7
`24.6
`
`predicted had essentially normal patterns of breathing, although
`the younger ones tended to be a bit tachypneic when breathing
`on the nebulizer. The child with the worse lung function (FEV1
`⬍ 30% predicted) was tachypneic at rest. The respiratory wave-
`forms were broken into 5-ms epochs and were used to calculate
`the expected deposition. Three breaths were chosen from a
`pattern that showed regular respiration, and the same three
`breaths were used to calculate expected deposition for each
`apparatus. Entrained flow was calculated by subtracting the
`nebulizer driving flow from the inspiratory flow. When this
`resulted in a negative number it was defined as zero, since the
`one-way inspiratory valve would be closed. The spring-loaded
`valve on the AeroEclipse does not open until the entrained flow
`reaches 8 L/min; output was considered zero until this occurred.
`From the algorithms of the total rate of output and that in the RF
`for the Pari LC Star and the AeroEclipse, the output in each 5
`ms-epoch for the specific entrained flow of the epoch was
`calculated and summed over the entire breath. These calculations
`are illustrated in Appendix 2. The results are reported as the
`mean of three breaths for each patient. This allows the in vivo
`efficiency, defined as the output during inspiration in the RF
`divided by total output over the entire respiratory cycle,8 to be
`calculated. For the Halolite, in vivo efficiency was equal to the
`output in the RF during inspiration since there is no expiratory
`drug loss.
`
`Validation of Assumptions
`
`To test the assumption that the output of the Pari LC Star and
`the AeroEclipse that was determined under steady-state condi-
`tions were valid under dynamic conditions, they were connected
`to the Harvard pump with the settings described above and run
`for 3 min. Total drug output (Ot) was calculated as described
`above. The two half waves from the Harvard pump were known
`mathematically and were entered as the “patient’s” breathing
`pattern. Using the algorithm for rate of drug output, the output
`over 3 min was calculated and compared to the measured output.
`Device evaluation and comparison included the expected
`pulmonary drug deposition per breath and per minute, in vivo
`efficiency, overall efficiency in terms of expected deposition in
`relation to the initial charge in the nebulizer, and for the Halolite
`the accuracy of the device to deliver a preset amount of drug. The
`calculated output and expected pulmonary deposition of the Pari
`LC Star and AeroEclipse, as well as the length of time to run to
`dryness were compared to the Halolite. To have comparable data,
`the time to deliver a selected predetermined dose was calculated
`for each device. The predetermined dose was defined as the dose
`delivered by the Halolite after four button presses, which was
`
`found to be essentially equivalent to its point of dryness. The time
`difference between the devices for each of the seven patients was
`calculated. The results are expressed as means ⫾ 95% confidence
`limits. Differences in patient size and device performance were
`explored by regression analysis.
`
`Results
`
`The steady-state in vitro assessment of both the
`total rate of output and that in the RF for the Pari
`LC Star and AeroEclipse is shown in Figure 4. With
`increasing entrained flow, the Pari LC Star increases
`both the Ot and that in the RF. The AeroEclipse
`begins producing aerosol when the entrained flow
`reaches 8 L/min (patient inspiratory flow is 13 L/min
`when the compressor driving flow is taken into
`consideration), and there is a slight fall off in Ot with
`increasing entrained flow, but there is an initial small
`increase in the RF, indicating a smaller particle size
`distribution with increasing flows. Given the design,
`the Halolite provides a constant output of 0.0029 mg
`per breath when it is activated.
`When the mathematically predicted output of the
`Pari LC Star and the AeroEclipse for the two half
`sinusoidal waveforms for the Harvard ventilator are
`compared to the actual output, there is no difference
`between the two (0.0089 ⫾ 0.0001 mg per breath
`vs 0.0090 ⫾ 0.0000 mg per breath, and 0.0046 ⫾
`0.0001 mg per breath vs 0.0045 ⫾ 0.0002 mg per
`breath for the Pari LC Star and the AeroEclipse,
`respectively [mean ⫾ 95% confidence index (CI)].
`
`Figure 4. Rate of output (total and in RF) in relation to
`entrained flow for the Pari LC Star and AeroEclipse while being
`driven by the compressor.
`
`www.chestjournal.org
`
`CHEST / 126 / 5 /NOVEMBER, 2004
`
`1623
`
`Liquidia's Exhibit 1103
`Page 5
`
`

`

`This gives credibility to the use of the mathematical
`model derived from steady-state data for the predic-
`tion of drug deposition during dynamic conditions.
`In other words, the quadratic equations that charac-
`terize the device performance during steady-state
`conditions can be applied during dynamic in vivo
`conditions.
`The rate of deposition as a function of respira-
`tory rate (Fig 5) is greatest for the Pari LC Star; it
`is intermediate for the AeroEclipse, and not re-
`lated to respiratory rate. In contrast, the Halolite
`has the lowest rate of deposition with most respi-
`ratory rates, but shows a linear increase with
`increasing rates until ⬎ 40 breaths/min, where the
`expected deposition is comparable to the other
`two devices. As expected, the opposite is true with
`tidal volume since those subjects with the greatest
`lung disease had the highest respiratory rates and
`the lowest tidal volumes. The mean rate of depo-
`sition ⫾ 95% CI was highest for the Pari LC Star
`(0.093 ⫾ 0.0084 mg/min with compressor
`vs
`0.120 ⫾ 0.0122 mg/min for dry air),
`lowest for
`the Halolite (0.055 ⫾ 0.016 mg/min, compressor
`only), and the AeroEclipse is in between (0.075 ⫾
`0.0064 mg/min with compressor, and 0.108 ⫾
`0.014 mg/min for dry air). The difference between
`the Pari LC Star and the AeroEclipse was signif-
`icant, but there was variability among the patients,
`resulting in overlap in the 95% CI between the
`AeroEclipse and the Halolite. The in vivo efficien-
`cies of Pari LC Star and AeroEclipse range from
`51 to 55% and 71 to 77%, respectively, for the
`older group of children. For the younger group,
`
`Figure 5. Rate of deposition as a function of respiratory rate for
`the Pari LC Star, AeroEclipse, and Halolite.
`
`the in vivo efficiencies are found to be 52 to 54%
`and 68 to 73%. The Halolite has an RF of 80%,
`which is equivalent to the in vivo efficiency. For
`the breath-actuated devices, the in vivo efficiency
`was essentially the RF since no drug is lost during
`expiration. For the Halolite, this is constant and
`independent of the subject, which is not the case
`for the AeroEclipse since increasing entrained
`flow played a role by increasing the RF (Fig 4). If
`the devices are run to dryness or “four button
`presses” for the Halolite,
`total expected drug
`deposition is greatest for the AeroEclipse (1.3032 ⫾
`0.0296 mg, compressor; 1.4719 ⫾ 0.0204 mg, dry
`air), least for the Halolite (0.8400 ⫾ 0.0000 mg), and
`intermediate for the Pari LC Star (0.9421 ⫾ 0.0307
`mg, compressor; 0.9719 ⫾ 0.0395 mg, dry air) and all
`independent of size of the subject. There is virtually
`no variation (95% CIs ⬍ 0.00005 mg of the initial
`dose of 2.5 mg of albuterol) in the expected dose
`delivered by the Halolite, despite large differences in
`size and breathing patterns. There was no relation-
`ship between the rate of deposition and the size of
`the subject, either in height or in weight. The larger
`(taller) subjects would have received less drug on a
`milligram per kilogram basis than the smaller sub-
`jects. This was most pronounced for the Aero-
`Eclipse, in which the 115-cm-tall subject would have
`received almost three times the amount in milli-
`grams per kilogram body weight than the subject 174
`cm in height. Furthermore, dosing differences due
`to device performance are greatest for the smaller
`subjects with much less discrepancy for the larger
`ones. With increasing entrained flow, the output and
`deposition are also increased, but the rate of Ot
`starts to level off at approximately 20 L/min for the
`AeroEclipse and approximately 30 L/min for the Pari
`LC Star.
`When evaluating the devices in terms of time to
`deliver a dose of medication, the Halolite consis-
`tently results in an expected pulmonary deposition of
`0.8400 mg with four button presses, which is there-
`fore selected as the comparing dose. Using the
`compressor, the Pari LC Star is fastest with expected
`pulmonary delivery in 9.2 ⫾ 0.8 min (mean ⫾ 95%
`CI) with the AeroEclipse taking 2.2 ⫾ 0.4 min
`longer, and the Halolite requiring 8.0 ⫾ 4.2 min
`longer. The much larger CIs with the Halolite are
`explained by the relationship of rate of output and
`respiratory rate with this device, whereas the other
`two devices are much less dependent on the respi-
`ratory pattern of the child. When driving the devices
`with hospital dry compressed air, the Pari LC Star
`delivers the dose in 7.1 ⫾ 0.7 min, with the Aero-
`Eclipse requiring 0.9 ⫾ 0.6 min longer.
`
`1624
`
`Laboratory and Animal Investigations
`
`Liquidia's Exhibit 1103
`Page 6
`
`

`

`Discussion
`
`This study demonstrates that three of the new
`generation of nebulizers each has particular strengths
`and weaknesses. In terms of rapid drug delivery, a
`factor that may shorten treatment time and improve
`adherence with recommended therapy in a disease
`like CF,16,17 the Pari LC Star appears to be the
`superior device when using a compressor. If the
`nebulizers are driven at 8 L/min from a compressed
`air source, as would be likely in a hospital setting, the
`rate of output for the Pari LC Star increases 29%,
`but 44% for the AeroEclipse, making the perfor-
`mance equivalent for the two devices. In terms of
`maximizing drug delivery, an important factor if the
`drug is very expensive, the AeroEclipse is the supe-
`rior device. From the perspective of a drug with a
`narrow therapeutic safety margin, the Halolite is
`much more predictable for drug delivery. Regardless
`of device, if a specific dose based on milligrams of
`drug per kilogram of body weight is desired, the
`initial dose put into the nebulizer will have to be
`individualized for the size of the patient. If the drug
`being nebulized is an antibiotic and minimal envi-
`ronmental contamination is desired, neither the
`AeroEclipse nor the Halolite allows significant (or
`any) antibiotic to leave the device except that which
`is exhaled by the patient.
`There are some potential limitations to this study.
`The most obvious is that in vitro data are combined
`with in vivo respiratory patterns to estimate, as
`opposed to measure,13 pulmonary deposition. While
`there is no doubt that nuclear medicine techniques
`would have to be considered the “gold standard,”
`comparison studies would mean multiple exposures
`to radioactive material that would not be allowed
`under the current regulations for ethical research in
`children. The comparisons are estimates, and the
`following issues are the potential sources of error in
`the calculations. The first is whether or not output
`data derived under steady-state conditions can be
`applied to the dynamic situation of regular breath-
`ing. In terms of total rate of output, the agreement
`between the data generated by the modified Harvard
`pump, which produces perfect half-sine waves, and
`the mathematical data using the model of the device
`output coupled to the mathematical expression of
`the sine waves suggests that steady-state data could
`be applied with no loss of accuracy. The second issue
`that is not addressed in this study is whether or not
`particle size distribution measured under steady-
`state conditions would apply to the dynamic situa-
`tion. In a previous study13 using the Pari LC Jet
`nebulizer in normal adults, very close agreement was
`found between the RF measured by laser diffraction
`and the in vivo RF measured by scintigraphy. How-
`
`ever, in children, the definition of an RF as the mass
`of aerosol carried in particles ⱕ 5 ␮m could be
`questioned. Recalculating the data of Wildhaber
`et al,22 it would appear that in smaller children a
`definition of ⱕ 5 ␮m for the RF is too large. Support
`for this comes from Geller and colleagues.23 who did
`not find evidence that smaller children received
`more tobramycin per kilogram of body weight from
`a Pari LC Jet nebulizer. They suggested that the
`lower RF in the small children limited pulmonary
`deposition. Since no specific data exist to give a valid
`estimate of RF based on size, this is a potential
`limitation that can be acknowledged but not scien-
`tifically corrected. Finally, the breaths used in the
`mathematical model assume a stable breathing pat-
`tern, by both the choice of where on the ventilatory
`pattern that the representative breaths are chosen,
`and the expression of the results as a mean of the
`three representative breaths. Such a stable pattern is
`frequently not the case in children. The device that
`would be most affected by an irregular pattern would
`be the Halolite because the timing of the output
`pulse is based on the previous three breaths. In a
`situation when breathing is irregular, the intended
`delivery of the pulse during the first half of inspira-
`tion may be mistimed. However, as a comparative
`study, the performance of each nebulizer is calcu-
`lated on the same three breaths from each patient,
`thereby minimizing physiologic variations in device
`performance. Another theoretical issue is that the
`manufacturer of the Halolite suggests that a better
`distribution of drug will occur if the pulse is deliv-
`ered only during the first part of inspiration, as
`compared to throughout the inspiratory cycle for the
`Pari LC Star and whenever inspiratory flow is ⬎ 13
`L/m in the case of the AeroEclipse. There is no
`specific comparative data to support or refute this
`claim, but a comparison between the Halolite and
`the Pari LC Jet, a less efficient precursor of the Pari
`LC Star,24 did not support more uniform distribution
`in patients with CF.
`There are differences in the in vitro performance
`between the Pari LC Star and the AeroEclipse. As in
`all breath-enhanced nebulizers, the entrained flow
`does two things: one is the reduction of rainout,
`thereby increasing output during the inspiratory
`phase; and the other is increasing the flow around
`the baffles, thereby increasing inertial impaction of
`the large droplets and reducing the particle size
`distribution. For the Pari LC Star, the predominant
`factor is the reduction of rainout so both the Ot and
`that in the RF increases with increasing entrained
`flow. In contrast, the Ot of the AeroEclipse fell,
`although the amount in the RF increases initially and
`then falls (Fig 4). As a result, the Pari LC Star is
`more efficient at rapid drug delivery than the Aero-
`
`www.chestjournal.org
`
`CHEST / 126 / 5 /NOVEMBER, 2004
`
`1625
`
`Liquidia's Exhibit 1103
`Page 7
`
`

`

`Eclipse; and both, due to their breath-enhanced
`design, are much faster than the unvented Halolite.
`In conclusion, each of the three devices tested has
`strengths and limitations. Both the choice of the device
`and the amount of drug placed in it should be made on
`the basis of the targeted pulmonary dose sought cou-
`pled with the desire for rapid delivery, the minimiza-
`tion of drug waste, and the need for precision of the
`pulmonary dose delivered. Hence,
`for prescribing
`medication to be given by aerosol, the initial dose, the
`device, and the patient must all be considered together
`if predicable results