Content uploaded by Michael McPeck
Author content
All content in this area was uploaded by Michael McPeck on Jun 04, 2022
Content may be subject to copyright.
REVIEW
Continuous infusion aerosol delivery of prostacyclins during mechanical ventilation:
challenges, limitations, and recent advances
Michael McPeck and Gerald C. Smaldone
Division of Pulmonary, Critical Care and Sleep Medicine, Department of Medicine, Stony Brook University Medical Center, Stony Brook, NY, USA
ABSTRACT
Introduction: Critically ill mechanically ventilated patients routinely receive aerosol delivery of epo-
prostenol by continuous infusion of the nebulizer by syringe pump. This procedure is ‘off-label’ as no
FDA approved drug presently exists. Without standardized protocols, therapy is based on prior experi-
ence with bronchodilators, limited studies of delivery systems and anecdotal clinical trials. Current
protocols based upon patient body weight and drug concentration determines the infusion rate of drug
dose delivered to the nebulizer , which is only distantly related to dose delivered to the lung and may
be altered by many factors.
Areas covered: This paper reviews the background of this technique as well as current methods of
managing drug delivery, technical challenges, and limitations. A recent advance in aerosol laboratory
bench testing, using radiolabeled aerosols, is presented to reveal important factors defining delivery.
Expert opinion: Off-label use of continuously nebulized prostacyclin in the ICU lacks the support of
large clinical trials needed for FDA clearance. However, comprehensive bench studies afford the
potential for clinicians to better understand and manage therapy at a level above simple dosing of
the nebulizer by body weight. New research techniques are enhancing our basic comprehension of the
interaction between aerosol devices and the mechanical ventilator.
ARTICLE HISTORY
Received 4 January 2022
Accepted 30 March 2022
KEYWORDS
Aerosol delivery; continuous
drug delivery; inhalation
administration; mechanical
ventilation
1. Introduction
Prostacyclins, when intermittently or continuously delivered
to the respiratory mucosa as an aerosol, exert selective
effects on pulmonary vasomotor tone, thereby reducing
local pulmonary vascular resistance without adversely influ-
encing the systemic circulation and are thus useful for treat-
ing pulmonary arterial hypertension (PAH) and ventilation/
perfusion mismatch in acute respiratory distress syndrome
(ARDS) [1–4]. Unlike other methods of drug delivery, inhala-
tion delivery of aerosols, especially during mechanical venti-
lation, can be challenging to implement with respect to
control of delivered dose and management of the various
technologies comprising the delivery system. Currently, in
the critical care setting, a new mode of aerosol delivery
during mechanical ventilation, referred to as ‘continuous
infusion aerosol delivery,’ [5] has evolved in which the nebu-
lizer is continuously infused with drug by a programmable IV
infusion pump to maintain continuous aerosol inhalation for
hours or many days, as clinically appropriate.
Prior experience with continuous short-acting β2 agonist
bronchodilators serves as the prototype for contemporary con-
tinuous infusion aerosol delivery of prostacyclins and is referred
to herein for that reason. Historically, the continuous nebuliza-
tion paradigm was first established in the late-1980s and 1990s
to provide extended delivery times of aerosolized β2 agonists
by face mask in pediatric and adult patients for treatment of
severe, reversible airways disease [6–15]. Continuously nebu-
lized β2 agonists were included in National Heart Lung Blood
Institute asthma guidelines as early as 1997 as a suggestion for
severe asthmatic patients that did not respond to routine
dosages of aerosol bronchodilators [16]. Clinical experience
has confirmed the safety of the technique when appropriate
monitoring (heart rate, ECG, and blood pressure) was con-
ducted, as is typical for patients receiving continuous
bronchodilators [17] as well as prostacyclins. The inhaled
dose is also defined by the output rate of the continuous
medication nebulizers, which are standardized. Based in part
on prior experience gained with continuous infusion aerosol
delivery of bronchodilators during mechanical ventilation,
this new mode of continuous nebulization has also become
an accepted procedure for epoprostenol aerosol delivery by
contemporary respiratory therapy services in the United
States, although its regulatory status is ‘off label.’
However, in the home care setting, two aerosolized pros-
tacyclin (PGI
2
) mimetics are currently FDA-approved for
chronic therapy for patients with advanced PAH: treprostinil
(Tyvaso® United Therapeutics Corporation) and iloprost
(Ventavis® Bayer Schering Pharma AG). Both agents were
FDA approved for administration by proprietary FDA cleared
electronic aerosol delivery systems [18,19] that permit con-
trolled treatments by mouthpiece in spontaneously breathing
patients. The delivery systems control the inhaled dose by
CONTACT Michael McPeck michael.mcpeck@stonybrookmedicine.edu Division of Pulmonary, Critical Care & Sleep Medicine, Stony Brook University
Medical Center, 100 Nicolls Road/T17-040 Health Sciences Center, Stony Brook, NY 11794-3869, USA
Supplemental data for this article can be accessed here
EXPERT OPINION ON DRUG DELIVERY
https://doi.org/10.1080/17425247.2022.2061460
© 2022 Informa UK Limited, trading as Taylor & Francis Group
detecting the patient’s breathing pattern with sensors and
then emit pulsed amounts of aerosolized drug to achieve
the prescribed inhaled dose. These are currently the only
prostacyclin drugs and associated delivery devices with regu-
latory approval in the United States.
For continuous mechanical ventilation in the ICU, no FDA-
approved drug or FDAcleared aerosol delivery device exists spe-
cifically for an aerosolized prostacyclin at the present time.
Nevertheless, epoprostenol sodium (Flolan® GlaxoSmithKline or
Veletri® Janssen Pharmaceuticals), formulated for intravenous
delivery, is now routinely nebulized ‘off label’ continuously dur-
ing mechanical ventilation for patients with refractory hypoxe-
mia associated with ARDS (Adult Respiratory Distress Syndrome).
This application has evolved, in part, from prior experience with
continuously inhaled bronchodilators, as well as the commercial
availability of a ‘continuous infusion kit’ and accompanying user
instructions for theSolo’ electronic vibrating mesh nebulizer
(VMN) that was brought onto worldwide markets by Aerogen
(Dangan, Galway, Ireland). Continuous aerosolized epoprostenol
usage has also been prompted by concerns about the high cost
borne by institutions that provide inhaled nitric oxide gas (iNO),
also a selective pulmonary vasodilator. Various institutional pro-
tocols have been developed to use aerosol epoprostenol to
accelerate weaning from iNO, or to start patients on aerosol
epoprostenol a priori, and transition them to iNO only if they
fail to respond to aerosol epoprostenol.
Contemporary continuous infusion aerosol delivery during
mechanical ventilation is a technologically complex proce-
dure employing specialized equipment [5]. The nebulizer is
replenished continuously by a calibrated infusion pump and
delivers its aerosol output into the heated and humidified
ventilator circuit in a manner that enables a portion of the
nebulizer output to be inhaled by the patient. The portion of
aerosol drug inhaled is termed Inhaled Mass, IM, and is
expressed as the percentage of drug mass delivered to the
nebulizer with respect to time. Traditionally, drug dose deliv-
ered in this manner may also be expressed as the actual drug
mass (in mg or µg) delivered during a specific time. However,
also expressing it as a percentage of the volume infused into
the nebulizer provides a measure of the efficiency of the
complete drug delivery system.
Albuterol dosing, which has not changed since the initial
protocols established in the 1990s, is simply prepared in four
different concentrations in volumes sufficient to deliver a mass
of 5, 10, 15 or 20 mg/h to the nebulizer for several hours and is
not adjusted for body weight, even for children. The infusion
pump is set for a constant volumetric flow not exceeding the
volumetric output rate of the nebulizer (so that the nebulizer
neither completely empties nor overfills).
Epoprostenol, conversely, is administered on a patient body
weight (PBW) basis with typical dosages ranging between 15
and 30 µg/kg/min infused into the nebulizer. When drug
amount (mg) and diluent volume (mL) in the syringe, plus the
PBW (kg) is entered into the programmable infusion pump, it
calculates and displays the dose to be delivered (µg/kg/min)
and the pump flow (mL/h) that will be provided to the nebu-
lizer. Continuous infusion aerosol delivery thereby allows main-
tenance of steady-state dose delivery to the nebulizer as well as
up or down titration of the delivered dose by manipulating the
infusion pump flow.
Despite the sophistication and precision of the infusion
pump system, several factors such as breathing pattern, pre-
sence of a humidifier, as well as the pump flow, all influence
the inhaled dose delivery (IM) to the patient to various
degrees [5]. Accordingly, the infusion rate of drug dose by
body weight delivered to the nebulizer by the pump (e.g. µg/
kg/min) is only distantly related to the actual dose delivered to
the airway opening or the lung.
1.1. Historical background
Aerosolized and inhaled prostacyclins have been studied
since at least 1933 for inducing pulmonary vasodilation in
patients with PAH, but initially little attention was paid to the
specifics of aerosol delivery technologies, which apparently
varied according to what was available or already in use at
the time. Appendix 1 presents a representative sample of
some of the early papers with comments pertaining to the
technology.
The first detailed in vitro bench study assessing the tech-
nology and performance characteristics of an infusion-based
aerosolized prostacyclin delivery system came in 2003. Siobal,
et al., described a clinical system for continuous aerosolized
prostacyclin therapy during mechanical ventilation with
a MiniHEART 30-mL jet nebulizer (SunMed, Tucson, Arizona)
at 2 L/min operating flow (8 mL/h output) [20]. The nebulizer
outlet was tee’d into the inspiratory limb of the ventilator
circuit close to the patient-Y and continuously replenished
Article highlights
Continuously aerosolized epoprostenol is routinely administered ‘off
label’ into the mechanical ventilator circuit in patients with pulmon-
ary arterial hypertension (PAH) and arterial hypoxemia due to Adult
Respiratory Distress Syndrome by continuously infusing the nebulizer
with the drug.
Continuous infusion aerosol delivery in critically ill intubated and
ventilated patients evolved from continuous infusion bronchodilator
delivery during mechanical ventilation that served as the prototype
for inhaled epoprostenol and which, in turn, evolved from continuous
nebulization delivery of bronchodilators by face mask in adult and
children with severe reversible airways disease.
Despite numerous clinical studies of aerosolized epoprostenol deliv-
ery by continuous infusion, with varying degrees of success, little has
been published regarding assessment of the technical aspects and
performance of the delivery system.
Current methods of in vitro aerosol delivery assessment (inhaled
mass, aerosol particle size determination, drugs, and radioisotopes
as surrogate tracers) are useful, and have been used successfully, for
example, to demonstrate in vitro to in vivo correlation. They are,
however, time-consuming to conduct and results are not available
until after the experiment has concluded and final analyses have
been accomplished.
Recent advances in in vitro testing using radiolabeled saline in an
infusion pump have enabled a deeper understanding of the multiple
factors that influence aerosol delivery and suggest exciting new
possibilities for determining real-time aerosol delivery during techni-
cally challenging procedures. Behavior of the system, as well as any
anomalies that may occur, can be observed in real-time so that
adjustments can be made on the spot and, if desired, multiple test
conditions can be explored during a single experiment.
This box summarizes key points contained in the article.
2M. MCPECK AND G. C. SMALDONE
by two infusion pumps that were tee’d together into the
nebulizer. One pump supplied the medication to the nebulizer
while the other provided saline diluent. The total infusion rate
was maintained at a constant flow of 8 mL/h, to match the
specified nebulizer output, also 8 mL/h. Drug concentration
delivered to the nebulizer was adjusted by changing the ratio
of infusion pump flow of medication to diluent flow. This
system had been used as rescue therapy in 11 patients with
ARDS requiring mechanical ventilation. To validate the system
on the test bench, a mechanical ventilation model was used
through which albuterol sulfate inhalation solution, 0.83 mg/
mL, as a surrogate for epoprostenol sodium was nebulized for
5 min (no chemical assay for epoprostenol was available to the
authors at that time). Albuterol emitted from the nebulizer
was collected in a filter at the distal tip of the endotracheal
tube, filters were washed to elute the captured albuterol into
solution, and the albuterol mass was determined with ultra-
violet spectrophotometry. This method determined that the
inhaled dose to the distal ETT filter averaged 14% of the
nominal dose placed in the nebulizer and the MMAD was
determined to be 3.1 µm sampled at the distal tip of the
endotracheal tube of the bench model [20].
The Siobal method gained popularity, influenced by the
desire of hospitals to reduce costs due to a dependence on
inhaled nitric oxide (iNO) gas. Concurrently, the first Aerogen
VMN for mechanical ventilation, Aeroneb ‘Pro,’ was introduced
in June 2002 and early adopters first began using it for inter-
mittent bolus treatments with unit-dose bronchodilators and
then continuous bronchodilator aerosol delivery with a single
infusion pump connected to the nebulizer’s solution reservoir.
The VMN connected into the ventilator circuit with a provided
tee piece, as did a large variety of small volume pneumatic jet
medication nebulizers (SVNs) that were already established in
the market. The inlet to the medication reservoir allowed the
connection of standard IV tubing with luer connector from the
infusion pump syringe and a needle puncturing the silicone
cap over the reservoir. In 2011, Aerogen added a color-coded
infusion syringe and delivery tubing (‘continuous infusion kit’)
that interfaced to the Aerogen ‘Solo’ VMN and published the
‘drop-by-drop’ continuous infusion technique instructions [21].
Convenient to use, and with wide company support in
a market with little to no competition, this method of contin-
uous infusion aerosol delivery gradually became the de facto
standard in respiratory therapy departments across the United
States due to its simplicity (only a single infusion pump
required) and ready compatibility with respiratory therapy
equipment.
In 2017, Anderson and colleagues conducted a series of
bench experiments that used ultraviolet spectrophotometry to
directly measure the mass of epoprostenol captured on a filter at
the distal tip of an ETT in a realistic critical care ventilator config-
uration set for a 50-kg patient model [22]. An Aerogen Solo VMN
was used as the aerosol generator and a programmable infusion
pump supplied epoprostenol to the VMN in three different flows
from a 60-mL syringe with an epoprostenol concentration of
15,000 ng/mL during each 20-min test run. Three breathing
patterns (all with approximately the same minute volume,
but different inspiratory duty cycles) were used. The goal of
their study was to determine the effect of nebulizer position in
the circuit on inhaled aerosol, i.e. close to the patient airway
or close to the humidifier. In this process, they also studied
nebulizer positioning on both dry (inlet) and wet (outlet) sides
of the heated humidifier plus two other positions close to the
airway opening.
Their results indicated no difference in drug delivery
between nebulizer placement on the dry side or wet side of
the humidifier. Drug delivery in both humidifier mounting
locations was essentially equal and both positions were
superior to VMN placement in the inspiratory limb close to
the patient, or between the patient-Y piece and ETT, even
closer to the patient. In addition, they reported a positive
correlation between epoprostenol dosing rate (ng/kg/min)
and epoprostenol mass captured on the filter. Although not
identified as such at the time, their data also indicated that
the delivery during different dosage scenarios were essen-
tially a function of the infusion rate of drug into the nebuli-
zer, as the concentration of epoprostenol in the infusion
syringe remained fixed. In other words, aerosol drug delivery
during continuous infusion into the nebulizer was directly
related to infusion pump flow.
2. In vitro assessment of continuous aerosol
delivery
2.1. Current methods
The clinical relevance and necessity of bench testing aerosol
delivery devices and protocols has never been more crucial
than with the application of continuously nebulized pulmonary
vasodilators to intubated patients receiving continuous mechan-
ical ventilation. There is no reliable way of predicting inhaled
aerosol based simply upon the mass and volume of medication
placed into a nebulizer or the concentration of drug prepared in
a syringe for continuous infusion nebulization.
Aerosol bench testing has necessarily evolved along with
the nebulizer technologies and is now capable of elucidating
not only the inhaled aerosol delivery, but also the ultimate
disposition of all medication introduced into the nebulizer to
account for inefficiencies in nebulizer systems, presence or
absence of a heated humidifier in the ventilator circuit, loca-
tion of the nebulizer in the circuit, the ventilator circuit tubing,
the airway interface (ETT or tracheostomy tube) and the var-
ious breathing patterns and ventilator settings that may be
used in contemporary critical care. Ideally, aerosol bench test-
ing should be designed to render in vitro data about system
performance that can predict in vivo delivery. To accomplish
that, a comprehensive understanding of the behavior of the
complete system is necessary, not just the nebulizer.
2.2. Inhaled mass
The amount of a drug delivered to the intubated and mechani-
cally ventilated patient is best represented by a measurement of
Inhaled Mass (IM), which can be defined in vivo as the mass of
aerosol drug (mg, µg, ng) that has transited the system and
exited the distal tip of the ETT and was thus delivered to the
patient during the inspiratory phase of the breathing cycle [23].
Once delivered in this context, further study of pulmonary
EXPERT OPINION ON DRUG DELIVERY 3
deposition and distribution of aerosol throughout the lung
requires in vivo testing. On the test bench, IM measured
in vitro is defined as the mass of aerosolized drug that deposits
on a collection filter, also usually placed at the distal tip of the ETT
of the ventilated model (Figure 1), and therefore represents the
drug that would have been inhaled if not intercepted by the
collection filter [23]. The experiment may be paused, and the
collection filter can be removed at any time for the measurement
of the aerosol that has accumulated on it. Some investigators
have used a single filter that rendered a single IM data point at
the end of a relatively short experiment [24]. For longer or more
complex experiments, multiple filters can be used at intervals,
and their measurements summed cumulatively to represent
a lengthy test run. Data points representing filter measurements
at intervals during an experiment can be plotted against time to
render an intuitively informative graph representing the rate of
drug delivery, the pattern of performance of the delivery system
and the IM as a function of treatment time [23].
A useful alternative nomenclature for expressing Inhaled
Mass is to identify the mass collected on the IM filter as
a percentage of the mass of drug placed into the nebulizer
(IM%). (The mass of drug placed into the nebulizer is often
called nominal dose, loading dose, or nebulizer charge). In this
manner, nebulizer or system efficiency can be expressed for
different conditions, e.g. different drug concentrations, load-
ing doses, bolus volumes, treatment times, ventilator breath-
ing patterns, infusion pump flows and so forth. Think of IM%
as a performance factor. Once the IM% of a specified aerosol
delivery system is established, actual drug mass delivery can
be readily calculated for any nominal dose. For example, if
a given aerosol delivery system is found to have an IM% of
20%, a treatment with a single 3 mL bolus of unit dose
albuterol 2.5 mg (the nominal dose), would render a mass
delivery of 0.5 mg (20% of 2.5 mg). A delivery system with
and IM% of 40% would deliver twice the drug mass if the
nominal dose was the same.
2.3. Aerosol Particle Size Distribution (APSD)
Another aspect of aerosol delivery involves the distribution of
particle sizes in the aerosol being studied. Nebulizers for
inhaled medication delivery produce polydisperse aerosols
with a range of largely spheroid particles of different aerody-
namic diameters, although the majority are typically below 10
micrometers (µm) in size due to nebulizer design. There is no
practical and convenient way of absolutely measuring the size
of the particles. Instead, using a cascade impactor device [26],
(or other techniques such as laser diffraction [27], which indir-
ectly estimate MMAD), the distribution of particle sizes across
the actual output range is inferred statistically and typically
expressed as a median value, i.e. the particle size in µm at 50%
of the total distribution. This is referred to as the Mass Median
Aerodynamic Diameter (MMAD), the diameter at which half
the particles are larger, and half are smaller.
For bench studies involving aerosol delivery during mechan-
ical ventilation, it is especially important to realize that the
MMAD of the aerosol stream emitted from the nebulizer may
be larger than that delivered to the patient at the distal end of
Figure 1. Schematic diagram 25, used with permission of authors] of the continuous infusion experimental setup showing the nebulizers on the dry side of the
humidifier. An infusion pump instilled radiolabeled solution into the nebulizer. In preliminary experiments [5], the IM filter, shown at the distal tip of the ETT, was
measured on the gamma camera at the conclusion of a 60-min experiment. In later experiments, a shielded gamma detector connected to a rate meter was added
and oriented to read the radioactivity accumulating on the IM filter at the distal tip of the ETT in real-time whenever triggered by the investigator. This configuration
also allowed experiments up to 4 hours in length by changing out IM filters at approximately 80–90 min intervals as they became saturated with condensed water
vapor from the ventilator’s humidification system.
4M. MCPECK AND G. C. SMALDONE
the circuit [28]. Therefore, APSD may be sampled at two different
locations. If sampled at the distal tip of the endotracheal tube,
where IM% is also determined, the MMAD is representative of
the aerosol mass being inhaled by the patient and is typically
smaller than the MMAD sampled directly at the nebulizer outlet.
This is because of circuit losses wherein the generally larger
aerosol particles in the distribution tend to ‘rain out’ prematurely
by various mechanisms, especially direct impaction on surfaces
in the delivery tubing, humidifier, and some of the connectors.
Conversely, if the APSD is sampled at the nebulizer outlet, where
a larger MMAD is most likely present, the difference in MMAD
between the two sampling sites may inform as to the cause and
nature of aerosol losses in the circuit. This is further complicated
by a heated humidifier device in the ventilator system, which
also decreases aerosol delivery [29]. Humidification, addition of
water in the vapor phase aided by warming the inspired gas to
approximately body temperature (37°C), tends to diminish the
MMAD at the distal tip of the ETT. The mechanism responsible is
hygroscopic growth in the ventilator circuit, wherein the particles
enlarge to some degree due to the condensation of water from
the available vapor, and then deposit in the tubing circuit leaving
the smaller particles to reach the airway opening.
2.4. Albuterol aerosol tracer
Theoretically, the aerosol drug of interest could be collected on
the IM filter and analyzed by a suitable chemical method.
Alternatively, some investigators have used surrogate aerosols
of unit dose albuterol sulfate inhalation solution (0.83 mg/mL),
for which readily available assays exist [22,24]. The IM of aerosol
collected on the filter can be determined by ultraviolet spectro-
photometry of albuterol that has been deliberately eluted from
the collection filter. Albuterol aerosol as a surrogate method is
usually sufficient for the determination of IM in a simple system.
Its disadvantage, due to measurement difficulties, is that it is
challenging to provide any determination of the aerosol deposi-
tion that occurs in various parts of a complete mechanical
ventilator system circuit (connectors, humidifier, inspiratory
and expiratory limbs, patient-Y piece, closed system suction
device (if used), and the ETT itself. This is mainly due to the
difficulty in recovering deposited albuterol from these areas for
measurement. Thus, accurate characterization of the complete
delivery system (Mass Balance) is difficult to achieve with only
an albuterol test aerosol. For that purpose, a radiolabeled sur-
rogate with physicochemical properties similar to inhalation
drugs can be measured by a suitable gamma radiation detector
in any part of the system in which the radioaerosol deposits [5].
2.5. Radiolabeled aerosol tracer
To expedite device testing and permit more detailed investi-
gation into the disposition of aerosol within a ventilator sys-
tem, radiolabeled normal saline or albuterol solutions may
be used as a tracer and/or a surrogate for other drugs
[5,12,25,29]. The optimal radiolabel for this application is tech-
netium pertechnetate (
99m
Tc), a gamma-emitter in a weak salt
solution not unlike normal saline or albuterol into which it is
mixed to obtain a working solution for charging a nebulizer
for study. For simple solutions, the radioactive label is uni-
formly dissolved in the solution and the mass of each neb-
ulized particle is proportional to the radioactivity in the
particle. As little radioactivity as 30 mCi of
99m
Tc is
a sufficient daily supply for most aerosol studies. For a short
nebulizer run of no more than an hour of nebulization, a few
drops (500–1,000 µCi or 0.5–1 mCi) added to the test drug or
normal saline solution is a safe and sufficient nebulizer dose.
The use of
99m
Tc as an aerosol tracer requires the ability to
measure its radioactivity in nebulizers, filters, and ventilator
circuit components. A radioisotope calibrator (e.g. Atom Lab
100, Biodex, Inc., Shirley, New York, or equivalent) may be
used for measuring activity in syringes, small nebulizers, and
filters that can fit into its measuring well. For more sophisti-
cated experiments, a gamma camera can measure nebulizers,
filters, and tubing and provide imaging of distribution of
radioaerosol in those circuit components. Technetium
99m
has
a half-life of 6 hours, the decay rate of which is exponential
and easily calculated in a spreadsheet.
2.6. In vitro to in vivo correlation
A comprehensive bench study in 2003, designed with atten-
tion to many of the experimental details mentioned pre-
viously, including the use of a radioactive tracer, serves as an
example of how detailed and nuanced in vitro data may
enable reasonable in vivo drug delivery prediction. Miller
et al., conducted in vitro aerosol delivery experiments with
three different critical care ventilators that represented the
technology of that time [26]. The experimental protocols
included two different brands of pneumatic jet nebulizer posi-
tioned in the inspiratory limb 30 cm from the patient-Y piece.
Breath actuation (inspiratory phase only) or continuous neb-
ulization modes were tested using relevant controls built into
the ventilator. Identical test runs were conducted with and
without a heated humidifier. Albuterol inhalation solution as
a 3-mL bolus radiolabeled with
99m
Tc was used as the nebu-
lizer solution. Nebulizer and filter measurements of radioactiv-
ity were performed with a radioisotope calibrator and APSD
with a low-flow cascade impactor. Parameters measured for
the different variables were: MMAD, IM%, and partial Mass
Balance (nebulizer residual plus sum of IM and exhalation
port collection filters).
The in vivo phase of the study was conducted on intubated
and mechanically ventilated patients enrolled in an IRB-
approval inhaled antibiotic protocol. Comparisons of inhaled
antibiotic delivery during breath-actuated vs continuous neb-
ulization and during humidified vs non-humidified ventilation
were conducted. Sputum was collected by tracheal suctioning
without saline instillation 1-h following aerosol therapy and at
8-h intervals for 24-h during which time sputum was once
again collected for analysis of sample weight and antibiotic
content. This study found that the concentration of drug in
the sputum was controlled by the nebulizer, the use of breath
actuation (nebulizer operating only during inspiration), and
the presence of humidity. Failure to control these parameters
resulted in limited control of drug delivery.
EXPERT OPINION ON DRUG DELIVERY 5
3. Recent advances in assessing continuous aerosol
delivery
3.1. Radiolabeled infusion syringe
In early 2021, our group applied the concept of radiolabeling
to the infusion syringe solution that was being infused into
the nebulizer and published the results of a bench model and
experimental protocol for studying continuous infusion aero-
sol delivery during adult mechanical ventilation over a period
of 1 hour [5]. Bench studies of continuous nebulization require
the ability to conduct prolonged aerosol delivery experiments
of 1–4 hours. For prolonged experiments in which a 60 mL
infusion pump syringe may be run for up to 4 hours, 5–6 mCi
of
99m
Tc-saline may be used in a single experiment to provide
a sufficient concentration of radioactivity and to accommo-
date the 6-h half-life of the isotope.
We also applied the principles of mass balance to continuous
nebulization to measure and image radioactivity deposited in
various parts of the delivery system with the gamma camera. This
allowed determination of the impact of the ventilator circuit,
humidifier, breathing pattern, nebulizer infusion flow, and even
the nebulizer on the delivered aerosol. The protocol relied upon
a radio-labeled infusion syringe as the basis for calculating drug
delivery and compared two nebulizers, an Aerogen Solo VMN
positioned on the dry side of the humidifier and a 3D-printed
(resin) prototype breath-enhanced jet nebulizer (BEJN) posi-
tioned on the wet-side (i-AIRE, InspiRx, Inc., Somerset, New
Jersey). Use of two different nebulizer technologies allows
a broader view of how devices and aerosols behave under
different circumstances, to aide in interpreting results.
The major variables in our initial protocol were the two
different nebulizer technologies (BEJN and VMN), the infusion
rate (1.5–12 mL/h) and the ventilator’s breathing pattern as
represented by the inspiratory duty cycles, 0.13–0.34 (13–34%
inspiratory time). The study’s goal was to define the factors
that influence continuous infusion aerosol delivery with an eye
toward developing control over the inhaled aerosol, much in
the same way as the Miller, et al. [26] study demonstrated.
While Inhaled Mass was still collected on a single filter at
the distal tip of the ETT as shown in Figure 1, a new technique
was applied for radiolabeling to account for the fact that the
nebulizer was charged by continuous infusion rather than
bolus instillation and was being replenished over time by
the infusion pump. The 60 mL volume of saline in the infusion
syringe was radiolabeled with
99m
Tc and measured by gamma
camera prior to the start of nebulization. A programmable
infusion pump instilled radiolabeled solution into the nebuli-
zer. In preliminary experiments [5], the IM filter, shown at the
distal tip of the ETT, was measured on the gamma camera at
the conclusion of a 60-min experiment. In current experi-
ments, the IM filter radioactivity accumulation was measured
as a rate rather than a finite amount: i.e. filter activity data
were presented as a percentage of the infusion syringe’s
original radioactivity per min, with appropriate corrections
applied to account for radioactive decay over time. To accom-
plish this, a shielded gamma detector connected to
a ratemeter was added and oriented to read the radioactivity
accumulating on the IM filter at the distal tip of the ETT in real-
time whenever triggered by the investigator. This configura-
tion also allowed experiments up to 4 hours in length by
changing out IM filters at approximately 80–90 min intervals
as they became saturated with condensed water vapor from
the ventilator’s humidification system.
Data were analyzed using multiple linear regression to
determine the contribution of each of the variables (nebulizer
type, duty cycle, and infusion flow) to the Inhaled Mass. The
key findings of this study were that nebulizer type (VMN vs
BEJN) was not an important factor with respect to Inhaled
Mass. Importantly, infusion pump flow was responsible for
87% of the variability of inhaled mass, and duty cycle an
additional 5% [5].
3.2. Real-time ratemeter and gamma camera
assessment
The methodology outlined above, refined to permit real-time
assessment of the radioaerosol accumulating on the Inhaled
Mass filter [25], reveals transient effects on the system that
would be lost when relying only on analysis of the IM filter at
the end of the experiment, for example, effects of changes in
ventilator settings or device malfunction. The current mechan-
ical ventilation model is also shown in Figure 1. The model
shown includes the adult critical care ventilator, circuit, and
test lungs, the prototype nebulizer configured on the ‘dry’
inlet side of the humidifier, and the IM filter positioned at
the distal tip of the ETT prior to the test lung. New to the
bench model, a shielded gamma scintillation detector inter-
faced to a ratemeter was positioned near the inhaled mass
collection filter to measure its accumulating activity in real
time. The shield and detector orientation minimized exposure
to other sources of radioactivity in the system.
During an experiment, one-minute readouts of accumu-
lated activity (counts/min) can be triggered as frequently as
desired to meet the goals of the protocol; for example, every
2 minutes for the first 10 minutes of a test run (as the nebu-
lizer output stabilizes after the start of infusion), and then
every 5 minutes thereafter, after it has reached a steady
state, or as further changes in test parameters are applied.
Each 1-min count replaces a single measurement of an Inhaled
Mass filter by other means. The investigator can also trigger an
ad hoc reading that provides instantaneous feedback about
nebulizer performance or potential anomalies in the experi-
ment; again, in real-time rather than at the end of an experi-
ment when the opportunity to assess or correct an irregularity
has passed. By entering each measurement into a spreadsheet
as it becomes available, an Inhaled Mass graph against time
can be plotted simultaneously (Figure 2).
Aerosol delivery is readily visualized using the real-time rate-
meter technique. The measurement system is sufficiently sensi-
tive to detect deviations from the steady-state as demonstrated
by deliberate interruptions and dynamic changes of the test
conditions during a 4-hr experiment [25]. As shown in Figure 3,
deliberately interrupting, or changing the parameters of the
infusion pump, the nebulizer, and/or the ventilator settings
6M. MCPECK AND G. C. SMALDONE
creates multiple test conditions to explore during a single test
run, as well as the ability of the real-time ratemeter technique to
detect anomalies. Intentionally interrupting infusion flow, nebu-
lizer gas flow, or electrical connection simulated an infusion
pump failure and nebulizer failure, respectively, to demonstrate
the characteristic patterns on the real-time results spreadsheet.
For clinical relevance, infusion pump flows and ventilator duty
cycle settings were changed to reproduce conditions that may
change in a clinical situation (Figure 3).
In conjunction with real-time ratemeter measurements, we
employed a gamma camera that is calibrated against the radio-
isotope calibrator on every test run in parallel with the initial
measurement of the nebulizer charge in microcuries (µCi). All
IM filters in a test run are also measured on the gamma camera
to calibrate and convert the ratemeter counts to a µCi standard
for the final measurements. Images are simultaneously made of
all components of the complete ventilator and aerosol delivery
system where radiolabeled aerosol has deposited prior to and
during the test run. In this manner, not only is the nebulizer
charge determined, but the radioactive aerosol deposited in, or
delivered to, all components in the ventilator circuit up to, and
including the IM filter. For some components, such as connec-
tors, tubing, and the humidifier chamber, a gamma camera
image may be valuable for determining the sites of deposition
and distribution of deposition within the component part
(Figure 4). In addition to imagin the deposition patterns of
aerosol within the components of the delivery system, measur-
ing each component allows calculation of the Mass Balance,
accounting for the radioactive solution infused into the nebu-
lizer during the test run (Figure 5). Finally, as necessary, repre-
sentative aerosol particle size determinations may be done on
different configurations of the test setup.
Figure 2. Two continuous infusion aerosol delivery experiments demonstrating
the value of real-time ratemeter assessment of aerosol delivery systems under
different conditions. Graph of ratemeter counts demonstrates the effect of
priming vs not priming the nebulizer prior to the start of a 60-min continuous
infusion test run. Closed squares denote the IM with a 2 mL prime of radi-
olabeled solution. A steep slope is shown for the first 5-min of operation during
which the 2 mL radiolabeled solution is nebulized. Thereafter, the nebulizer
achieves a steady state consistent with the infusion rate, with a relatively
constant slope indicating a steady aerosol delivery rate to the IM filter. Closed
squares show ratemeter counts done at 1-min intervals for the first 13-min of an
otherwise identical experiment without a nebulizer prime. Only background
radioactivity is initially detected. At 15-min the system achieves a steady state
with a slope like the previous slope for the primed nebulizer [25], used with
permission of authors].
Figure 3. A continuous infusion aerosol delivery experiment in which anomalies
were deliberately created to assess the sensitivity of the real-time ratemeter to
detect them. At event 1 the gas flow to the pneumatic nebulizer is turned off to
simulate an accidental disconnect. No aerosol output is present until the gas
flow is restored at event 2, at which time the accumulated radioactivity is
nebulized with a greater output until the original steady state (same slope) is
restored at event 3. An identical tracing would occur for an electronic VMN in
which electrical power was accidentally disconnected. At event 4 the infusion
pump was turned off, simulating an accidental disconnect. Aerosol output
quickly stops until the pump is turned back on again at event 5. In this case,
the slope of the output is the same as the steady state prior to the anomaly, as
no radiolabeled solution could accumulate in the nebulizer while the infusion
pump was turned off. [25, used with permission of authors].
EXPERT OPINION ON DRUG DELIVERY 7
Open
4. Conclusion
Continuous infusion aerosol delivery has evolved into a complex
but readily applicable procedure for management of PAH and
ARDS in the critically ill. The use of radiolabeled aerosols and
a bench model that closely duplicates actual clinical conditions,
and provides information and feedback in real time, provides
a more complete understanding of the delivery of these aerosols
in less time and with greater detail than a single measurement of
an IM filter. The use of continuous infusion aerosol delivery is
increasing in the intensive care unit with recognition that it is
a valuable drug delivery technology during invasive ventilation. It
has the potential to displace other methods of inhalation therapy
with application during noninvasive ventilation and high-flow
nasal oxygen therapy. This evolution requires the continuing gui-
dance of comprehensive in vitro aerosol delivery testing to max-
imize the potential benefits that can be derived from this therapy.
Figure 4. Example gamma camera images of components parts of the complete ventilator system demonstrating the distribution of aerosol deposition across the
system. Inspection and analysis of these images provide information about extent of aerosol losses in the circuit that diminish the inhaled mass. For example,
deposition in the inspiratory circuit limb is shown with a ‘hot spot’ at the tubing inlet due to turbulence at the transition from the humidifier outlet to the tubing.
25, used with permission of authors].
Figure 5. Representative Mass Balance data from a single experiment, shown graphically, demonstrates the distribution of radioactivity throughout the bench
model at the conclusion of a test run. The total recovery (98.1%) of the radioactivity infused into the nebulizer during the test run, indicates a technically acceptable
experiment. The nebulizer residual (35.5%) is radioactivity that remained in the nebulizer. The amount of radioactivity that was emitted from the nebulizer is
distributed throughout the different components of the ventilator circuit as shown in Figure 4. The ‘inhaled dose’ delivered to the ‘patient’ (IM Filter) is 39.2% of the
total amount of the radioactivity infused into the nebulizer during the 90-min test run. 30, used with permission of authors].
8M. MCPECK AND G. C. SMALDONE
5. Expert opinion
Continuous infusion aerosol delivery of epoprostenol during
mechanical ventilation, allows precise control of targeted ther-
apy to the lungs in critically ill patients with ARDS. The worldwide
presence of the vibrating mesh nebulizer with its infusion syringe
generated broad interest among intensivists primarily as an
alternative to inhaled nitric oxide, resulting in widespread use
during the COVID-19 pandemic. Despite the adoption of contin-
uous infusion aerosol delivery by VMN among respiratory care
and critical care practitioners, there is no standardized therapeu-
tic protocol established by clinical trials. Instead, practitioners,
some with the help of their hospital pharmacy department,
devised protocols from information that was circulated via the
professional grapevine: for example, online forums where
respiratory care practitioners exchange policies and procedures,
manufacturer’s instructions for use, company sales representa-
tives, and presentations and exhibits by users and company
personnel at professional meetings.
Off-label treatments and protocols for important clinical
problems are a natural part of medical practice. For safety,
and to predict efficacy, there should be research studies that
provide evidence to support the use of new methods.
Studies of continuous infusion nebulization are limited.
Understanding the delivery of aerosolized medications begins
with the principle that the dose of drug placed in the nebulizer is
only distantly related to the dose delivered to the lung. The
delivered dose is affected by the ventilator circuit, its settings,
nebulizer function, and the patient’s physiology. These factors
are best studied on the bench before attempting clinical trials.
The benefits to be realized from this approach include under-
standing: (1) the range of possible drug delivery from low to high
doses, (2) factors of the ventilator circuit that affect delivery in
real time as settings are changed, and (3) the reliability of devices
over time (continuous nebulization can used for hours or days).
This review summarizes recent work that illustrates techniques
for defining immediate changes in aerosol delivery as some of
these parameters are changed during an experiment.
Rigorous testing on the bench facilitates comparison between
different commercial devices as well as new prototypes in devel-
opment. Clinicians cannot rely solely on the FDA 510(K) medical
device clearance-to-market process to answer broad questions of
drug delivery in clinically relevant situations. Economic aspects
can be important. A major driver that leads to adoption of con-
tinuous aerosolized epoprostenol inhalation was an effort to offset
the high cost of inhaled nitric oxide (iNO) gas, which is also
a selective inhaled pulmonary vasodilator. However, aerosol deliv-
ery is more complex than delivery of the gas. Nitric oxide is
monitored in real time in the circuit. This is not possible clinically
with aerosol therapy, but it can be done on the bench with
radioactive tracers as demonstrated with real time ratemeter mea-
surements [25]. Economic costs of devices can be significant.
Enhanced bench testing could be employed to study other
devices and technologies that could potentially achieve the
same treatment objectives while reducing operating costs.
An area deserving increased scrutiny involves the determina-
tion of the nominal dose of drug that is infused into the nebu-
lizer. Currently, the algorithm programmed into infusion pumps
for the delivery of epoprostenol to a continuous nebulizer in
a ventilator circuit requires that the clinician enter the patient’s
ideal body weight (IBW) in kg derived from standard anthropo-
metric tables into the infusion pump, as well as the dose to be
delivered as a rate (mcg/kg/min). For a given drug concentration
in the infusion syringe, the infusion pump will calculate and set
the appropriate infusion rate (mL/h) necessary to achieve the
desired dose rate. The range of dosing should be set by an
understanding of device performance on the ventilator and the
patient should define the response. By understanding the func-
tion of the delivery system, the clinician should be assured of
drug delivery and confidently increase the dose until there is
a response or side effect.This approach is essentially how con-
tinuous albuterol inhalation has been managed since the late
1980s: four standardized delivery rate increments (5, 10, 15, and
20 mg/h) are generally used, while the patient is monitored for
response and potential side effects [12]. Because epoprostenol,
like albuterol, is a short-acting agent, the key to this approach is
physiological monitoring, which is typically already being done
by continuous acquisition of oxygen saturation by pulse oxime-
try and vital signs such as pulse rate, electrocardiogram, arterial
blood pressure and, frequently, pulmonary artery pressure).
Careful dose titration guided by physiological monitoring should
render the desired clinical response. Bench studies could deter-
mine upper and lower delivery limits that would ensure dose
control and safety [25].
Declaration of interest
G Smaldone serves as a consultant to InspiRx and is a member of the
advisory board. Stony Brook University holds patents on nebulizers
licensed to InspiRx. The authors have no other relevant affiliations or
financial involvement with any organization or entity with a financial
interest in or financial conflict with the subject matter or materials dis-
cussed in the manuscript apart from those disclosed.
Funding
This paper was not funded.
Reviewer disclosures
Peer reviewers in this manuscript have no relevant financial or other
relationships to disclose.
ORCID
Michael McPeck http://orcid.org/0000-0003-4632-8573
Gerald C. Smaldone http://orcid.org/0000-0002-3798-4942
References
Papers of special note have been highlighted as either of interest (•) or of
considerable interest (••) to readers.
1. Dzierba AL, Abel EE, Buckley MS, et al. A review of inhaled nitric
oxide and aerosolized epoprostenol in acute lung injury or
acute respiratory distress syndrome. Pharmacotherapy. 2014;34
(3):279–290.
2. Searcy RJ, Morales JR, Ferreira JA, et al. The role of inhaled prosta-
cyclin in treating acute respiratory distress syndrome. Ther Adv
Respir Dis. 2015;9(6):302–312.
EXPERT OPINION ON DRUG DELIVERY 9
3. Pacheco J, Arnold H, Skrupky L, et al. Predictors of outcome in 216
subjects with ARDS treated with inhaled epoprostenol. Respir Care.
2014;59(8):1178–1185.
4. Torbic H, Szumita PM, Anger KE, et al. Inhaled epoprostenol vs
inhaled nitric oxide for refractory hypoxemia in critically ill patients.
J Crit Care. 2013;28(5):844–848.
5. McPeck M, Ashraf S, Cuccia AD, et al. Factors determining contin-
uous infusion aerosol delivery during mechanical ventilation. Respir
Care. 2021;66(4):573–581.
• First in vitro study using radiolabeled medication surrogate
and multiple regression analysis to compare different nebuli-
zer types and determine that continuous infusion aerosol
delivery is dependent on infusion pump flow rather than
nebulizer type.
6. Moler FW, Hurwitz ME, Custer JR. Improvement in clinical asthma
score and PaCO
2
in children with severe asthma treated with
continuously nebulized terbutaline. J Allergy Clin Immunol.
1988;81(6):1101–1109.
7. Portnoy J, Aggarwal J. Continuous terbutaline nebulization for
treatment of severe exacerbations of asthma in children. Ann
Allergy. 1988;60(4):368–371.
8. Voss KR, Willsie-Ediger SK, Pyszczynski DR, et al. Description of
a delivery method for continuously aerosolized albuterol in status
asthmaticus. J Asthma. 1990;27(1):37–39.
9. Colacone A, Wolkove N, Stern E, et al. Continuous nebulization of
albuterol (salbutamol) in acute asthma. Chest. 1990;97(3):693–697.
10. Chipps BE, Blackney DA, Black LE, et al. Vortran High Output Extended
Aerosol Respiratory Therapy (HEART) for delivery of continuously
nebulized terbutaline for the treatment of acute bronchospasm.
Pediatr Asthma Allergy Immunol. 1990;4(4):271–277.
11. Moler FW, Johnson CE, Van Laanen C, et al. Continuous versus
intermittent nebulized terbutaline: plasma levels and effects. Am
J Respir Crit Care Med. 1995;151(3_pt_1):602–606.
12. McPeck M, Hughes K, Tandon R, et al. Aerosol delivery during
continuous nebulization. Chest. 1997;111(5):1200–1205.
• Among the first bench studies to explore the continuous neb-
ulization paradigm and report that the early dosing protocols
(mg/hr nebulizer output) did not accurately reflect actual
inhaled drug delivery.
13. Raabe OG, Wong TM, Wong GB, et al. Continuous nebulization
therapy for asthma with aerosols of β2 agonists. Ann Allergy
Asthma Immunol. 1998;80(6):488–508.
14. Kelly HW, Keim KA, McWilliams BC. Comparison of two methods of
delivering continuously nebulized albuterol. Ann Pharmacother.
2003;37(1):23–26.
15. Berlinski A, Willis JR, Leisenring T. In vitro comparison of 4
large-volume nebulizers in 8 hours of continuous nebulization.
Respir Care. 2010;55(12):1671–1679.
16. NHLBI, National Asthma Education and Prevention Program. Expert
Panel Report II. Guidelines for the diagnosis and management of
asthma. Publication no. 97-4051. Bethesda MD: US Department of
Health and Human Services; 1997.
17. Khine H, Fuchs SM, Savile AL. Continuous vs intermittent nebulized
albuterol for emergency management of asthma. Acad Emerg Med.
1996;3(11):1019–1024.
18. Tyvaso Inhalation System. Instructions for Use Manual. [cited 2022
Feb 28]. Available from: https://www.tyvaso.com/pdf/TD300_
instructions_for_use.pdf
19. Tips to Help Make the Most of Your Treatment with VENTAVIS/
Mastering the ABCs of the I-neb Adaptive Aerosol Delivery (AAD®)
System. [cited 2022 Feb 28]. Available from: https://www.janssen
carepath.com/sites/www.janssencarepath-v1.com/files/ventavis-
patient-brochure.pdf
20. Siobal MS, Kallet RH, Pittet J-F, et al. Description and evaluation of
a delivery system for aerosolized prostacyclin. Respir Care. 2003;48
(8):742–753.
• First published paper with detailed description of a clinically
used continuous infusion aerosol delivery system plus an
in vitro assessment of its performance.
21. Aerogen Solo System Instructions. P/N 30-674 Rev L, 2020, 24–26.
[cited 2022 Feb 28]. Available from: https://www.aerogen.com/wp-
content/uploads/2020/09/30-674-Rev-L-Aerogen-Solo-System-
Instruction-Manual-US_WEB.pdf
22. Anderson AC, Dubosky MN, Fiorino KA, et al., The effect of nebu-
lizer position on aerosolized epoprostenol delivery in an adult lung
model. Respir Care. 62(11): 1387–1395. 2017.
• Contemporary bench study of common continuous infusion
aerosol delivery system comparing nebulizer mounting loca-
tions and measuring inhaled mass of epoprostenol.
23. Smaldone GC. Drug delivery via aerosol systems: concept of “aero-
sol inhaled.” J Aerosol Med. 1991;4(3):229–235.
24. Ari A, Atalay OT, Harwood R, et al. Influence of nebulizer type,
position, and bias flow on aerosol drug delivery in simulated
pediatric and adult lung models during mechanical ventilation.
Respir Care. 2010;55(7):845–851.
25. McPeck M, Lee JA, Cuccia AD, et al. Real-time In Vitro assessment of
aerosol delivery during mechanical ventilation. Journal of Aerosol
Medicine and Pulmonary Drug Delivery. 2022;35(1):32–40.
•• Pilot studies of bench measurements of radiolabeled aerosol
with gamma ratemeter demonstrates ability to obtain real-
time operational feedback and detailed analysis of factors
influencing aerosol delivery as the experiment in progressing,
rather than after the experiment has concluded. Steady states,
trends, and anomalies can be readily identified. Multiple test
conditions can be explored in a single experiment. Coupled
with gamma camera imaging, complete mass balance account-
ing for disposition of all aerosol in system can be readily
determined. Holds promise as a more versatile and efficient
method of conducting aerosol delivery studies in technically
difficult circumstances, such as continuous infusion aerosol
delivery and trans-nasal aerosol delivery by high flow nasal
cannula
26. Miller DD, Amin MM, Palmer LB, et al. Aerosol delivery and modern
mechanical ventilation: in vitro/in vivo evaluation. Am J Respir Crit
Care Med. 2003;168(10):1205–1209.
• In vitro-to-in vivo correlation study establishing that charac-
terizing behavior of an aerosol delivery system on the test
bench can be leveraged to enhance the control of inhaled
dose delivery to the lung.
27. Sidler-Moix A-L, Di Paolo ER, Dolci U, et al. Physicochemical aspects
and efficiency of albuterol nebulization: comparison of three aerosol
types in an in vitro pediatric model. Respir Care. 2015;60(1):38–46.
28. McDermott K, Oakley JG. Droplet size and distribution of nebulized
3% sodium chloride, albuterol, and epoprostenol by phase Doppler
particle analyzer. Curr Ther Res Clin Exp. 2021;82. doi:10.1016/j.
curtheres.2021.100623.
29. O’Riordan TG, Greco MJ, Perry RJ, et al. Nebulizer function during
mechanical ventilation. Am Rev Respir Dis. 1992;145(5):1117–1122.
30. Walmrath D, Schneider T, Pilch J, et al. Aerosolized prostacyclin in
adult respiratory distress syndrome. Lancet. 1993;342(8877):961–962.
10 M. MCPECK AND G. C. SMALDONE
Ref 30 should be Ref 1 in the Appendix
online supplement.
