Home Search Collections Journals About Contact us My IOPscience

Benchmarking the ERG valve tip and MRI Interventions Smart Flow neurocatheter

convection-enhanced delivery system's performance in a gel model of the brain: employing

infusion protocols proposed for gene therapy for Parkinson's disease

This article has been downloaded from IOPscience. Please scroll down to see the full text article.

2012 J. Neural Eng. 9 026009

(http://iopscience.iop.org/1741-2552/9/2/026009)

View the table of contents for this issue, or go to the journal homepage for more

Download details:
IP Address: 72.33.57.124
The article was downloaded on 30/01/2013 at 15:02

Please note that terms and conditions apply.

IOP PUBLISHING JOURNAL OF NEURAL ENGINEERING
J. Neural Eng. 9 (2012) 026009 (13pp) doi:10.1088/1741-2560/9/2/026009

Benchmarking the ERG valve tip and

MRI Interventions Smart Flow
neurocatheter convection-enhanced
delivery system’s performance in a gel
model of the brain: employing infusion
protocols proposed for gene therapy for
Parkinson’s disease
Karl Sillay 1,2 , Dominic Schomberg 1 , Angelica Hinchman 1 ,
Lauren Kumbier 1 , Chris Ross 3 , Ken Kubota 4 , Ethan Brodsky 1,2
and Gurwattan Miranpuri 1
1
Department of Neurological Surgery, University of Wisconsin, Madison, WI 53792, USA
2
Department of Biomedical Engineering, University of Wisconsin, Madison, WI 53706, USA
3
Engineering Resources Group Inc., Pembroke Pines, FL 33029, USA
4
Kinetics Foundation, Los Altos, CA 94023, USA
E-mail: sillay@neurosurgery.wisc.edu

Received 25 July 2011

Accepted for publication 21 December 2011
Published 14 February 2012
Online at stacks.iop.org/JNE/9/026009

Abstract
Convection-enhanced delivery (CED) is an advanced infusion technique used to deliver
therapeutic agents into the brain. CED has shown promise in recent clinical trials. Independent
verification of published parameters is warranted with benchmark testing of published
parameters in applicable models such as gel phantoms, ex vivo tissue and in vivo non-human
animal models to effectively inform planned and future clinical therapies. In the current study,
specific performance characteristics of two CED infusion catheter systems, such as backflow,
infusion cloud morphology, volume of distribution (mm3) versus the infused volume (mm3)
(Vd/Vi) ratios, rate of infusion (µl min−1) and pressure (mmHg), were examined to ensure
published performance standards for the ERG valve-tip (VT) catheter. We tested the
hypothesis that the ERG VT catheter with an infusion protocol of a steady
1 µl min−1 functionality is comparable to the newly FDA approved MRI Interventions Smart
Flow (SF) catheter with the UCSF infusion protocol in an agarose gel model. In the gel
phantom models, no significant difference was found in performance parameters between the
VT and SF catheter. We report, for the first time, such benchmark characteristics in CED
between these two otherwise similar single-end port VT with stylet and end-port non-stylet
infusion systems. Results of the current study in agarose gel models suggest that the
performance of the VT catheter is comparable to the SF catheter and warrants further
investigation as a tool in the armamentarium of CED techniques for eventual clinical use and
application.
(Some figures may appear in colour only in the online journal)

1741-2560/12/026009+13$33.00 1 © 2012 IOP Publishing Ltd Printed in the UK & the USA
J. Neural Eng. 9 (2012) 026009
1 Introduction
Convection-enhanced delivery (CED) is an advanced
technique used to distribute therapeutic agents into targeted
regions of the brain (Bobo et al 1994). In order for fluid
convection to occur, pressure gradients must be maintained
during the infusion process. This is typically accomplished
through the use of mechanically controlled infusion pumps and
has been found to have a penetrative advantage compared to
diffusion-driven administration (Morrison et al 1994). Studies
have found this approach to be more efficient and clinically
useful as compared to simple diffusion-driven administration
as it produces a larger and more precise volume of delivered
agent over a shorter period of time (Ratliff and Oldfield
2001). CED also produces a more evenly distributed and larger
volume of drug over diffusion-driven delivery. Because direct
intracranial administration of drugs can be delivered using this
method, larger molecules that could not otherwise penetrate
the blood–brain barrier can be introduced to the tissues of the
brain. This technique shows promise for therapeutic treatment
of a variety of medical conditions (Bankiewicz et al 2000,
Sampson et al 2007, Rogawski 2009).
While the applications of CED technology are promising,
improvements are still needed before it finds broad clinical
use. Previous clinical trials have noted the inadequate drug
distribution and the occurrence of backflow along the catheter–
brain interface (Richardson et al 2011). Both tissue damage
and low infusion rates have also been problematic in clinical
trials (White et al 2011). Thus, the development of new
catheters and infusion protocols is in demand.
In our experiments, we benchmark the ERG valve-
tip (VT) catheter (Engineering Resources Group, Pembroke
Pines, FL) against the Smart FlowTM (SF) catheter (MRI
Interventions, Memphis, TN). We also examine the best
practice infusion strategy as previously published by our
colleagues at the University of Wisconsin (Emborg et al
2010) and at the University of California at San Francisco
(Richardson et al 2011).
We conducted a series of agarose gel infusions with the
ERG and MRI Interventions infusion systems. Agarose gel
has an advantage over brain tissue when conducting physics
of infusion experiments due to the translucent nature of the gel
allowing us to capture the infusion morphology in real time
with high definition video. Agarose gel has been shown to
model a subset of characteristics of in vivo tissue relevant
to CED, such as volumes of distribution (Vd) and certain
gel–catheter interactions, and can be used as a reasonable first
step in evaluating CED protocols (Chen et al 2002). Agarose
gel of similar concentration to that used in the current study
has been reported to closely model the free diffusion value in
the interstitial space of the brain (Nicholson and Tao 1993).
Limitations of using argrose gel as a surrogate for in
vivo tissue are not insignificant. When inserting an infusion
catheter into gel, the gel structure may lacerate and fracture
unpredictably and even maneuvers such as catheter rotation
after a gel–catheter interface has occurred will alter the
catheter–gel interface in ways that are likely dissimilar to the
brain–catheter interface. The authors have reported limitations
2
K Sillay et al
to the gel model when working with larger molecules
(40 kDa) or significantly charged molecules, reducing the
diffusion coefficient by 5–70% depending on the magnitude
of these factors (Schantz and Lauffer 1962). Nicholson
and colleagues have suggested that variance in diffusion
characteristics found in gel may be more pronounced in brain
tissue.
We surmise that issues associated with molecular size
and charge do not pose significant limitations to the current
study as our chosen infusate is less than 1 kDa and carries
no significant charge. Another limitation may be in overall
infusion morphology. In the current study, spherical end
infusion morphology was common. In attempting to reproduce
these results in in vivo models, one must consider that brain
tissue is more anisotropic, heterogeneous and contains regional
anatomical boundaries. Despite these limitations, we believe
that agarose gel is a valid surrogate with respect to backflow,
infusion cloud morphology and Vd/Vi ratio changes during
infusion.
Specific performance parameters of the catheters were
examined to determine if published standards for the SF
catheter are possible with the VT catheter. The aim of
this work is to prove or disprove the ability to use the
catheters interchangeably with respect to published infusion
protocols, aiming device, advancing device and infusion
device substitutions.
The primary outcome in benchmarking is catheter
backflow. Secondary outcome measures include volume of
distribution versus volume of infusion (Vd/Vi) ratios, rate of
infusion, infusion cloud morphology and infusion pressure.
We report for the first time such benchmark characteristics
between infusion catheters in CED between otherwise similar
VT and end-port non-stylet catheters.
2. Methods
Gel infusion experiments were performed at the University of
Wisconsin, Madison, in the summer of 2011. A total of 63
infusion clouds were created in 26 unique agarose gels using
equipment and protocols as defined below.
2.1. Equipment
The major pieces of equipment used for the current study
include clear flow catheter (MRI Interventions, Irvine, CA),
VT catheter and pressure monitor (Engineering Resources
Group, Pembroke Pines, FL), Navigus trajectory guide
(Medtronic Corp, Minneapolis, MN) and PHD2000 MRI
syringe pump (Harvard Apparatus, Holliston, MA).
2.2. Catheter specifications
Two catheter designs were evaluated in this study: a VT
catheter with a retractable stylet and a single-endport catheter
without a stylet (SF).
The VT catheter used in our study was manufactured by
ERG in accordance with the specifications detailed below. The
catheter has a single-endport stepped design with a silica stylet
J. Neural Eng. 9 (2012) 026009
running within the entire length of the 3 ft tubing/catheter
apparatus. The fused silica shaft has an outside diameter
0.65 mm and an inner diameter 0.32 mm. The tip is polyimide
with an outer diameter 0.36 mm and an inner diameter
0.25 mm. The tip extends 3 mm from the end of the silica
shaft. The VT catheter employs a valve feature at the distal
tip to prevent end-port occlusion and tissue from entering the
catheter during placement. The VT was designed so as to allow
proper infusion function after being retracted a short distance
(approximately 5 mm) without being removed. It is not thought
to prevent infusate backflow.
The single-endport 16-gauge SF catheter without a stylet
has an outer diameter 1.65 mm and inner diameter 0.20 mm.
The central lumen is manufactured from non-reactive silica.
This catheter was used with two different line lengths, 4 and
10 ft, with respective priming volumes of 40 and 100 µl.
2.3. Agarose gel preparation
A volume of 1 l of 0.2% agarose gel solution was made
using 100 mL of 10 × Tris/borate/EDTA (TBE) buffer (Life
Science Products), 900 ml of deionized water and 2 g agarose
powder (Ultra Pure AquaPor LE). The solution was heated
until the agarose was completely dissolved (see appendix A).
Gel casings were then filled and allowed to cool and solidify
for approximately 1–2 h. After this brief cooling period,
approximately 1.5 cm of distilled water was added to the
surface of the gel. Cases were stored in a 4 ◦ C refrigerator
for a maximum of 4 days before use.
2.4. Infusate tracker preparation
Bromophenol blue, an odorless blue–black water-soluble dye
(MW 690), was chosen as a surrogate to track the distribution
of infusate during CED into the agarose gel (Chen et al
2004). It was prepared to provide sufficient visual contrast
between the infusate and agarose gel without precipitating in
solution.
2.5. Tubing preparation
Sections of 3 ft Teflon⃝R
perfluoroalkoxy tubing extension lines
(outer diameter 1.600 mm, inner diameter 0.051 mm) were cut
and attached to two SF Nuts 1/16, two flangless ferrules, one
M6 female to male luer assembly (Tefzel⃝ R
, ETFE, DuPontTM ),
one female luer to female union and one Teflon⃝ R
Tub
fluorinated ethylene propylene. The male luer assembly was
then connected to the catheter and the entire system was flushed
with distilled water until all air bubbles were removed. Infusate
was loaded into a 5 ml Hamilton syringe and mounted to a
programmable infusion pump (Harvard apparatus PHD 2000).
A 2 inch pressure monitor extension (ERG) was attached to
the syringe and tubing and connected to the computer. The
syringe, extension line and catheter tubing system were pre-
loaded and flushed with infusate in excess before beginning
the infusion process. The tip of each catheter was kept in saline
until the infusion is ready to begin.
3
K Sillay et al
2.6. Infusion computer-control specifications and line
preparation
Two programmable infusion/withdraw pumps capable of
advancing or retracting the plunger of attached Hamilton
syringes were synchronized with and fully controlled by
customized computer software (Lenovo ThinkPad with Multi-
channel Pressure Recorder v1.17.6, ERG, Inc.). Prior to
an experimental run, the syringe diameter and an infusion
script were input into the pump controller software. The
input infusion script controlled the rate and volume of
infusion throughout the entire infusion process. This capability
allowed for stepped ramp infusion protocols to be performed.
System line, infusion rate, infusion volume, and transduced
pressures and times were recorded and displayed using the
aforementioned software.
To ensure that saline had not diluted the infusate near the
catheter tip, infusate was once again run through each catheter
before being removed from the saline solution. The catheter
mounts were then transferred to the gel container tops. The
catheter was inserted by hand through guides on the mounts
until the tip was suspended in the center of the saline solution
just above the gel surface. The pressure was zeroed and the
catheter insertion was performed, according to the infusion
protocol (appendix B).
2.7. Catheter preparation and insertion technique
The VT catheter was inserted into the gel using the remote
introducer. To minimize the risk of catheter endport occlusion,
the VT catheter employs a valve feature in the catheter tip.
A stylet inside the VT catheter was advanced just beyond the
tip to close the valve producing a barrier to prevent tissue
from entering the endport. Once advanced, the VT stylet was
retracted approximately 5 mm to open the valve. The line
pressure was then zeroed as above. After the VT catheter
had been properly positioned, the SF catheter was inserted
by hand to match the depth of the VT catheter. In order to
minimize the risk of end port occlusion with the SF catheter,
the positive pressure was maintained at 1.0 µl min−1 during
catheter advancement (personal communication, Russ Lonser
and Krys Bankiewicz). Catheter insertion occurred smoothly
over a period of approximately 5 s (see appendix B).
2.8. Agarose gel infusion
The infusion process was started simultaneously for both
catheters while the SF catheter was being inserted into the
gel and the VT catheter was already in place. Once the final
programmed infused volume was reached, the infusion pumps
automatically stopped.
In the event, additional infusions were desired in the same
gel, the catheter was advanced at least 15 mm and the above
protocol was repeated. The infusion script was restarted and
the infusion program was run again. Advancement protocol
followed the same procedure as insertion protocol. This was
repeated up to three additional times per gel casing. After the
final infusion, at least 10 min elapsed before removing the
catheters from the gel (see appendix B).
J. Neural Eng. 9 (2012) 026009 K Sillay et al

2.9. Infusion protocols

The UCSF group and UW group both reported on preclinical
infusion protocols, which differed significantly. The UCSF
group proposed the infusion protocols for an upcoming AAV2-
GDNF human clinical trial. These protocols were used as
benchmarks for the reported experiments (Richardson et al
2011, Emborg et al 2010). Each of the following three
techniques for infusions was performed in the current study:
(a) UW 1.0 C—continuous infusion at 1.0 µl min−1 over
25 min for a total volume of 25 µl;
(b) UCSF 3.0 R—stepped ramp protocol in which the infusion
rate increased from 1.0 to 3.0 µl min−1 in 0.5 µl steps
every 5 min over 25 min for a total volume infused of
50 µl;
(c) UCSF 5.0 R—stepped ramp protocol in which the infusion
rate increased from 1.0 to 5.0 µl min−1 in 1.0 µl steps
every 5 min over 25 min for a total volume infused of Figure 1. Method of the backflow analysis. If the measured distance
75 µl. traveled by the infusate along the catheter–gel interface is greater
than the distance traveled by the infusate through the gel distal to the
catheter tip, backflow is considered to be detectable and is measured
2.10. Data collection from the catheter tip to the proximal catheter–infusate border.
Pressure measurements were read by the ERG pressure
monitor and recorded by the computer software. Video was cloud characterization.) The inclusion criteria for a spherical
recorded throughout the entire infusion process using a high- cloud were met if the difference between the height and width
definition (HD) video camera (Canon Vixia HF M31 HD). of the cloud was less than or equal to 20%. Inclusion criteria
Snap shots were taken each time the rate of infusion was for an ellipsoidal cloud were met if the difference between the
changed and approximately every 5 min for 25 min after the height and width of the cloud was between 20% and 40%. All
final infusion volume was reached. other clouds were characterized as irregular. This analysis was
performed on each cloud at the time the infusion ended.
Relative ranges are as follows:
2.11. Data analysis
spherical: 1 < x ! 1.25 (0–20% difference)
For the purposes of this paper, we individually analyzed each
infusion cloud and performed statistical analysis as such. ellipsoidal: 1.25 < x ! 1.66 (20–40% difference)

irregular: x > 1.66 (>40% difference).

2.11.1. Backflow measurements. Backflow was measured
from the tip of the catheter to the point at which the dye stopped
traveling up the catheter (figure 1). Detectable backflow
was recorded if this distance was greater than the distance
traveled by the dye below the catheter tip. Backflow, when
detected, was found to begin at the time the infusion was
initiated and/or when the infusion rate was changed (see
stepped pressure tracings over time (figure 2)). In each case, a
backflow measurement was taken when infusate movement up
the catheter–gel interface had stopped, as determined by the
retrospective analysis of video data.
2.11.4. Volume calculation for spherical infusions. The
diameter was measured and divided by 2 in order
to approximate the radius of the cloud. Volume was
calculated using the standard formula for a sphere, (4/3)π r3
(figure 4, volume of distribution calculation). The volume of
each infusion cloud was calculated at 5 min intervals from
the beginning of the infusion to 25 min after the infusion had
ended. This calculation results in the volume of distribution
(Vd) for a spherical infusion cloud at any given time.

2.11.5. Volume calculations for non-spherical infusions.

2.11.2. Height/width ratios. Height and width measure-
ments were made (mm) for each infusion cloud. Height and
width measurements were expressed as ratios and used for
quantitative characterization infusion clouds.
2.11.3. Infusion cloud morphology. As the prototypical
infusion is spherical, an additional analysis of each infusion
cloud was performed to determine its morphology. Each cloud
was qualitatively characterized as spherical, ellipsoidal or
irregular using the height/width ratios. (See figure 3, infusion
In clouds not characterized as spherical, a multiple-slice
integration technique was used. Each infusion cloud was
divided into 10–15 equidistant horizontal (axial) slices.
The horizontal diameter of each slice was measured with
the assumption that a circular cylindrical slice was being
approximated. A single vertical measurement was also taken to
determine the infusion cloud height. A numerical integration
formula was then used to approximate the cloud volume
(figure 4). This calculation results in Vd for a non-spherical
infusion cloud at any given time.

4
J. Neural Eng. 9 (2012) 026009 K Sillay et al

(A)

(B)

Figure 2. Pressure versus time: line infusion pressure versus time is shown in two infusion experiments. In experiment (A), previously
reported in vivo putaminal infusion protocols were compared (UW 1.0 C for VT and UCSF 3.0 R for SF). In experiment (B), both catheters
were benchmarked using a proposed ramped protocol planned for a human infusion trial (UCSF 5.0 R for VT and SF). (A) VT catheter
infusing at 1.0 µl min−1 in four separate infusion clouds. (Note: pressure monitoring detected a protocol deviation in the second VT infusion
where the stylet was not retracted from the tip.) The SF catheter was used to deliver a ramped infusion rate protocol to 3.0 µl min−1 in
0.5 µl min−1 increments every 5 min. (B) Both VT and SF catheters infusing a ramped infusion protocol increasing from 1.0 to
5.0 µl min−1 in 1.0 µl min−1 increments every 5 min. Differences in line pressure are consistent with catheter inner diameter as well as
length according to Newtonian physics calculations. SF catheter line pressure is higher, as expected, for a given infusion rate. In all cases,
the lower pressure trace belongs to the VT catheter.

Figure 3. Infusion cloud characterization. Infusion clouds were characterized as spherical, ellipsoidal or irregular depending on the
calculated height/width ratio.

5
J. Neural Eng. 9 (2012) 026009
Figure 4. Volume of distribution calculation. Volume of distribution (Vd) was calculated based upon the measured diameter for spherical
infusions. For infusions not qualifying as spherical by height/width measurements, multiple cross-sectional measurements were obtained as
the basis for Vd calculations described above.
2.11.6. Vd/Vi ratios. The volume detected (Vd) is reported
as the calculated volume (mm3) in which bromophenol blue
was detected as present. The volume infused (Vi) is reported
as the volume (µl) delivered by the computerized infusion
pump system. The Vd/Vi ratio has units of mm3 µl−1 and is
reported to demonstrate the pore fraction of gel as compared
to that expected in tissue targeted for human clinical infusions.
Cubic millimeters per microliter alerts the reader to the fact
that infused bromophenol blue concentration may vary slightly
over the detected volume and this variation is not used in the
calculation of the total concentration of the infusate at any
given point.
3. Results
A total of 63 infusions were performed in 26 unique agarose
gels (appendix C). Of these, 31 infusions were performed with
the VT catheter and 32 infusions were performed with an
SF catheter. While every piece of available data was used in
our analysis, some video data were lost for a small portion
of experiments. Additionally, data from one experimental day
were excluded because of equipment damage. As a result,
the total infusion clouds analyzed may differ depending
on the analysis conducted. In our paper, infusion clouds
were analyzed for (1) detectable backflow and distance, (2)
morphology, (3) volume and (4) initial and maximum infusion
pressures.
6
K Sillay et al
Table 1. Observation of detectable backflow by the catheter and
protocol.
Protocol VT SF
UW 1.0 C (n = 14) 5 of 14 (35.7%) N/A
UCSF 3.0 R (n = 16) N/A 8 of 16 (50.0%)
UCSF 5.0 R (n = 20) 5 of 10 (50.0%) 6 of 10 (60.0%)
Total (n = 50) 10 of 24 (41.7%) 14 of 26 (53.8%)
Infusion clouds were analyzed for the presence of detectable
backflow. No statistically significant difference was found
between catheters (p = 0.3926) or infusion protocol used.
3.1. Backflow analysis
Backflow data were available for 50 infusion clouds. Backflow
was detected in 24 of 50 infusion clouds. Backflow was found
to occur in 10 of 24 VT infusions and 14 of 26 SF infusions.
The overall mean backflow distance was 8.89 ± 3.93 mm. No
statistically significant difference was found between backflow
incidences in the VT catheter versus the SF catheter (p =
0.3926). Comparison with respect to the UCSF 5.0 R protocol
did not show statistical significance (p = 0.6547) (table 1).
On two occasions, backflow was observed during a
change in the infusion rate in addition to the infusion start
point. In these cases, only the final backflow distance was
included in our analysis. In every other instance detectable
backflow occurred during the first stage of the infusion process
and ended before the rate of infusion increased beyond
1 µl min−1. Two categories of the mean backflow distance
were calculated. The first, comprehensive mean backflow, was
J. Neural Eng. 9 (2012) 026009 K Sillay et al

Table 2. Quantification of backflow by the infusion protocol and catheter.

VT VT SF SF
UW UCSF UCSF UCSF VT SF Overall
1.0 C 5.0 R 3.0 R 5.0 R total total total
Backflow 5 of 14 5 of 10 8 of 16 6 of 10 10 of 24 14 of 26 24 of 50
incidence (35.7%) (50%) (50.0%) (60.0%) (41.7%) (53.8%) (48.0%)
Comprehensive
mean 2.12 ± 2.95 4.52 ± 4.76 4.71 ± 4.86 6.30 ± 5.42 3.1 ± 4.20 5.3 ± 5.8 4.27 ± 4.49
backflow (mm) (n = 14) (n = 10) (n = 16) (n = 10) (n = 24) (n = 26) (n = 50)
Exclusive
mean 5.94 ± 1.43 9.04 ± 3.76 9.42 ± 3.91 10.5 ± 4.2 7.71 ± 3.17 10.51 ± 3.77 8.89 ± 3.93
backflow (mm) (n = 5) (n = 5) (n = 8) (n = 6) (n = 10) (n = 14) (n = 24)
Detectable backflow measurements were made and analyzed comprehensively, with all infusion clouds, and exclusively, only with
infusion clouds with detectable backflow. There was no significant difference between protocols when clouds were analyzed
comprehensively (VT protocols, p = 0.1921; SF protocols, p = 0.4688; catheter type, p = 0.1361). A statistical difference was
found on the analysis performed on only those clouds with detectable backflow. Backflow was greater for the VT UCSF 5.0 R
versus the VT UCSF 1.0 C protocol (p = 0.0049) as well as for the SF versus the VT catheter (p = 0.0091).

Table 3. Infusion cloud characterization compared with backflow.
No of infusion
Height/width clouds with
ratio range Characterization backflow
(1 < x ! 1.25) (n = 29) Sphere 5 of 29 (17%)
(1.25 < x ! 1.66) (n = 11) Ellipsoid 9 of 11 (82%)
(x > 1.66) (n = 10) Irregular 10 of 10 (100%)
Spherical infusion clouds exhibited significantly fewer infusions
with backflow than ellipsoidal infusion clouds (p = 0.0002) and
irregular infusion clouds (p = 0.0001). No statistically significant
difference was observed between infusion clouds characterized as
ellipsoids and infusion clouds characterized as irregular
(p < 0.2148).
found by including every infusion cloud in the mean backflow
distance calculation. If no backflow was observed, the distance
was 0 mm. No statistical difference was found between the
UW 1.0 C and UCSF 5.0 R protocols for the VT catheter
(p = 0.1921) nor for the UCSF 3.0 R and UCSF 5.0 R protocols
for the SF catheter (p = 0.4688) (table 2).
A subanalysis of the mean backflow distance, exclusive
mean backflow, was conducted for only infusion clouds with
detectable backflow. When separated by catheter type, average
VT-derived infusion cloud backflow was 7.71 ± 3.17 mm
(n = 10), while SF was 10.51 ± 3.77 mm (n = 14).
This distance was statistically significant (p = 0.0091). With
respect to different protocols, a mean backflow distance of
5.94 ± 1.43 mm (n = 5) for the 1.0 C protocol and 9.04 ±
3.76 mm (n = 5) for the 5.0 R protocol was determined for
the VT catheter and was shown to be statistically significant
(p = 0.0049). The mean backflow in the SF catheter was
9.42 ± 3.91 mm (n = 8) for the 3.0 R protocol and 10.5 ±
4.2 mm (n = 6) for the 5.0 R protocol and was not statistically
significant (p = 0.4339) (table 2).
3.2. Infusion cloud morphology
A total of 51 infusion clouds were analyzed for morphology
(table 5). Total prototypical spherical infusions were obtained
in 30 of 51 infusions. Morphology classification was also
Table 4. Height/width ratios compared with backflow.
Backflow (n = 50) Average height/width ratio
No (n = 24) 1.16 ± 0.098
Yes (n = 26) 1.63 ± 0.389
The total average sphere height was calculated
and compared with the presence of detectable
backflow. This difference was statistically
significant (p = 0.0001).
compared to the incidence of backflow (table 3). In summary,
5 of 29 infusion clouds characterized as spheres, 9 of 11
infusion clouds characterized as ellipsoids and 10 of 10
infusion clouds characterized as irregular also had observed
backflow. A statistically significant relationship with respect
to the incidence of backflow was seen between spherical
and ellipsoidal infusion clouds as well as between spherical
and irregular infusion clouds (p = 0.0002 and 0.0001).
However, no statistically significant difference was noted when
ellipsoidal infusion clouds were compared to irregular infusion
clouds (p <0.2148) (table 6).
Further analysis was performed to relate the height/width
ratio to the incidence of backflow. The average height/width
ratio of infusion clouds without observed backflow was 1.16
± 0.098 mm (n = 24), while that with observed backflow was
1.63 ± 0.39 mm (n = 26). This difference was statistically
significant (p <0.0001) (table 4). The average sphere height
was also compared to the incidence of backflow and was also
found to be statistically significant (p = 0.0016). Further
analysis on the average infusion cloud height/width ratio
with respect to catheter type was performed and no statistical
differences were found.
Subanalysis was performed on infusion clouds
characterized as spherical with respect to the catheter and
infusion protocol. The VT catheter produced a total 17 of
24 spherical infusion clouds, while the SF catheter produced
a total 13 of 27. Although a trend was observed, no statistical
difference was found between the catheter used and spherical
morphology (p = 0.1041). Similarly, no statistical significance
was found when comparing the performance of the VT catheter

7
J. Neural Eng. 9 (2012) 026009 K Sillay et al

Table 5. Spherical morphology compared with the protocol. Table 7. Maximum infusion pressures by catheter type and infusion
protocol.
Experiment VT SF
VT (3 ft) SF (4 ft) SF (10 ft)
UW 1.0 C 11 of 13 (84.6%) N/A Experiment (n = 26) (n = 19) (n = 6)
UCSF 3.0 R 1 of 1 (100%) 7 of 17 (41.2%)
UCSF 5.0 R 5 of 10 (50%) 6 of 10 (60%) UW 1.0 C 8.2 ± 6.9 mm Hg N/A N/A
Total (n = 51) 17 of 24 (70.8%) 13 of 27 (48%) (n = 15) (n = 15)
UCSF 3.0 R 6 mm Hg 39.7 ± 11.5 mm Hg N/A
Infusion cloud classified as spheres was compared (n = 16) (n = 1) (n = 15)
with the catheter and protocol used. No significant UCSF 5.0 R 12.2 ± 55.5 ± 108.7 ±
difference in the percentage of infusions categorized 4.2 mm Hg 6.7 mm Hg 7.0 mm Hg
as spherical was found in the analysis of catheter type (n = 20) (n = 10) (n = 4) (n = 6)
or infusion rate. (UCSF 3.0 R, p = 0.2531; UCSF
5.0 R, p = 0.6547; total, p = 0.1041.) Statistically significant differences were observed when
comparisons were made with respect to the catheter and protocol
Table 6. Vd/Vi ratios by the catheter and infusion protocol. used (p <0.05), except when UW 1.0 C was compared with UCSF
5.0 R for the VT catheter (p = 0.10).
Experiment VT SF
UW 1.0 C 5.0 ± 1.8 (n = 12) N/A Table 8. Average line pressure versus infusion rate with UCSF
UCSF 3.0 R 6.6 (n = 1) 5.4 ± 2.3 (n = 12) 5.0 R.
UCSF 5.0 R 4.2 ± 1.1 (n = 10) 4.8 ± 1.2 (n = 10) Rate VT 3 ft SF 4 ft
Total (n = 45) 4.7 ± 1.7 (n = 23) 5 ± 1.8 (n = 22)
1 µL min−1 3.6 15.1
Vd/Vi ratios were compared with the catheter and protocol 2 µL min−1 6.9 25.3
used. No statistically significant difference in Vd/Vi ratios 3 µL min−1 9.8 32.6
was found with respect to the catheter (p = 0.8226) or 4 µL min−1 13 40.3
protocol used. 5 µL min−1 16.1 49.8

to SF with respect to the protocol used (p = 0.2531 and 0.6547)

(table 6).
3.3. Vd/Vi ratios
Vd/Vi ratios were available for 45 infusions and an analysis
was performed with respect to the catheter and protocol used.
The overall average for these clouds was 4.96 ± 1.85. The
Vd/Vi ratio for the VT catheter was 4.7 ± 1.70 (n = 23) and
the overall Vd/Vi ratio for the SF catheter was 5.15 ± 2.01
(n = 22). The UW 1.0 C protocol with the VT catheter resulted
in a mean Vd/Vi ratio of 5.0 ± 1.8. The UCSF 3.0 R protocol
for the VT and the SF catheter resulted in a Vd/Vi ratio of
6.6 and 5.4 ± 2.3, respectively. The UCSF protocol (UCSF
5.0 R) with the VT catheter resulted in a mean Vd/Vi ratio of
4.2 ± 1.1, while the SV catheter resulted in a mean Vd/Vi
ratio of 4.8 ± 1.2. No significant differences were found when
comparing the protocol used and thecatheter (table 6).
3.4. Infusion pressure
An analysis of the maximum infusion pressure for 51 infusion
clouds was conducted. Of these, 26 were created with the VT
catheter and 25 with an SF catheter. Pressure recordings were
separated and compared by the catheter length and type as well
as by the protocol. The maximum infusion pressure differed
by a factor of 6.6 between the VT and SF 4 ft catheter with
respect to the UCSF 3.0 R protocol. The max infusion pressure
differed by a factor of 4.5 between the VT and SF 4 ft catheter
and by a factor 8.9 between the VT and SF 10 ft catheter with
respect to the UCSF 5.0 R protocol. The max pressure for the
UCSF 3.0 R and UCSF 5.0 R protocols with respect to the
catheter used differed by factors of 2.0 and 1.4, respectively.
The max pressure for SF 4 ft and SF 10 ft differed by a factor
of 2.0 with respect to the UCSF 5.0 R protocol (table 7).
According to the properties of Newtonian fluids, the
infusion line pressure is proportional to the flow rate and line
length given the consistency of other parameters π ∗ (pressure)
∗
(radius)4/8 ∗ (viscosity) ∗ (length). In these experiments, the
flow rate was modified as well as the line length. As expected,
increases in the flow rate caused an increased measured line
pressure in both the VT and SF catheter. These differences are
accounted for by internal volume and line length calculations.
The mean pressure readings for the VT and SF catheters
for two representative infusion runs are included as follows: a
pressure reading from an experiment benchmarking the UCSF
5.0 R protocol using the VT and SF 4 ft catheters and a pressure
reading using the same protocol using an SF 10 ft catheter.
Average line pressures during each infusion are displayed
(tables 8 and 9).
A comparison between the presence of an initial pressure
spike and backflow incidence was made for each infusion cloud
(data not included). The four categories were then analyzed
using a chi-square test. No statistical significance was found
when all four categories were compared (p = 0.1949).
4. Discussion
4.1. Gel and surrogate choice
Agarose gel phantoms were chosen as surrogates for our
experiments. Agarose gel has been shown to display similar
poroelastic properties as brain tissue, a property due to swelling
caused by infusate penetration into interstitial spaces. Agarose
gel at 0.2% has been shown to mimic the changes in local
pore fraction caused by gel dilation due to CED (Chen et al
2002). Additionally, the transparency of gel allows for data to
be easily taken from infusion clouds without the use of MRI or

8
J. Neural Eng. 9 (2012) 026009 K Sillay et al

Table 9. Average line pressure versus infusion rate with UCSF to say that backflow does not occur at higher infusion rates,
5.0 R. but only that we are not able to detect any backflow in these
Rate VT 3 ft SF 10 ft cases because the infusion cloud perimeter has surpassed the
distance backflow must travel in order to be detected. The
1 µl min−1 5.1 28.1
additional time that elapses before infusion rates are increased
2 µl min−1 3.9 40.6
3 µl min−1 5.5 60.8 in ramped protocols may also allow the catheter–gel interface
4 µl min−1 6.7 75.7 to form a tighter bond and thus prevent any additional backflow
5 µl min−1 8.3 96.2 from occurring.
Average line pressure was As only ramped infusions were used to obtain infusion
calculated for each infusion rate rates higher than 1 µl min−1, the question of the relationship
during the formation of a single between infusion cloud and higher flow rates in a ramped
infusion cloud. Statistically versus non-ramped protocol becomes apparent. As the current
significant differences were study was designed to test best practice infusion protocols and
observed when comparisons
were made with respect to
then compare performance of both catheters during ramped
catheter and protocol used (p infusions, no direct answer to the question of VT versus SF
<0.05), except when UW 1.0 C catheter backflow with a non-ramped 5 µl min−1 infusion can
was compared with UCSF 5.0 R be drawn. If there is an advantage to ramped protocols from a
for the VT catheter (p = 0.10). backflow standpoint in the future, certain advantages could be
obtained, including a higher final infusion rate which may lead
other technology. Bromophenol blue was chosen as a tracker to decreased infusion time for patients receiving CED infusion
of the infusate distribution due to its relatively low molecular therapy.
weight as compared to other dyes, such as trypan blue, and Aside from the infusion rate and catheter diameter, other
its precedent for use when testing CED in agarose gel (Panse factors known to influence backflow include shear modulus
et al 2010). and hydraulic conductivity (Raghavan et al 2010). These
factors can make direct backflow comparisons between gel
4.2. Backflow and brain tissue difficult.

No significant differences between backflow and catheter

type or protocol used were observed except when exclusively
analyzing the infusion clouds with detectable backflow. In this
case, backflow was found to be significantly greater when
protocols with higher infusion rates were used. It is possible
that changes in the infusion rate increased the probability of
occurrence of detectable backflow. Additionally, backflow was
found to be significantly greater when the SF catheter was used.
This may be due to the greater diameter of the SF catheter. The
catheter diameter has been shown to correspond to backflow
(Morrison et al 1999).
The backflow distance in the current study was measured
from the tip of the infusion catheter to the edge of detectable
backflow as in figure 1. Although some authors have subtracted
the radius at the distal end of the backflow from the distance
(Martin Brady, personal communication), we have chosen to
report the backflow only from the catheter tip to the edge
of detection to avoid confounding the measurements with
multiple measurements. The current measurement protocol
may overestimate backflow slightly and the results should be
taken into account accordingly.
Backflow has been associated with catheter geometry and
infusion rate (Raghavan et al 2006). As the infusion rate
increases, it is reasonable to infer that backflow would also
increase. However, our studies have shown backflow to occur
primarily during the first few minutes of the infusion process,
which was always started at 1 µl min−1. Further backflow may
occur at higher infusion rates; however, visualization of this
may be obscured by the expanding infusion cloud properly.
One hypothesis for the lack of detection of increased backflow
during transitions to higher flow rates is that the backflow
was less than the radius of the infusion cloud. This is not
4.3. Infusion cloud morphology
We report that 30 of 51 infusion clouds were of spherical
morphology according to our inclusion criteria. As the
spherical infusate distribution is ideal, this low frequency
was less than anticipated. It is possible that low concentration
agarose gel may encourage the unequal distribution of infusate
during CED; however, more in vitro studies would be needed
to confirm this hypothesis.
We present in this study a novel quantitative method of
determining infusion cloud morphology as described earlier.
Few studies have related infusion cloud morphology to
backflow. We have chosen to analyze cloud morphology
in terms of the height/width ratio to determine a range at
which the occurrence of backflow during an infusion can be
inferred. As shown in table 4, these ratios appear to correlate
with the incidence of detectable backflow. We have also
decided to categorize some non-prototypical infusion clouds
as ellipsoidal. Although many ellipsoidal infusion clouds
have also been shown to have backflow in our experiments,
infusions of this nature would often be acceptable in vivo
outcomes and are thus distinguished from a typical non-
spherical infusion cloud characterization.
4.4. Vd/Vi
Vd/Vi ratios relate primarily to the pore fraction of the
material. While these ratios may be similar, the mechanism
by which the infusate diffuses through the medium is not and
one should advise cautious interpretations. However, Vd/Vi
ratios can be used as a reference to validate our methods and
use of gel surrogates in preparation for in vivo and clinical

9
J. Neural Eng. 9 (2012) 026009
testing. Our Vd/Vi ratios are comparable to those observed
in 0.2% agarose gel (Chen et al 2002). However, the current
results differed from other reported infusions in agarose gel
of different concentrations. Recently, standard protocols for
infusion into non-human primate putamen were performed
and reported Vd/Vi ratios of 3.3 ± 0.3 (n = 17), significantly
different from our reported Vd/Vi ratio of 4.96 ± 1.85
(p < 0.0001) (Emborg et al 2010, Richardson et al 2011).
In the setting of prototypical infusion cloud morphology,
one would hypothesize equal Vd/Vi infusion calculations
given an identical gel and an identical infusion protocol. In
one set of experiments, it appears that there was a trend toward
different Vd/Vi ratios; however, the hypothesis matched our
experimental result in finding no significant difference in end-
infusion Vd/Vi between the VT and SF catheters.
Other limitations in Vd/Vi calculations in this study are
acknowledged. Volumes were hand segmented from recorded
HD video. Measurements by a single investigator were used to
limit inter-observer technique variations; however, this method
remains less objective than a computer-derived segmentation
method.
4.5. Pressure
In our experiments, we compare two clinical infusion protocols
as well as benchmark a clinical infusion protocol between
two different catheters (figures 2(A) and (B)). As indicated
by these pressure readings, changes in the infusion rate are
associated with pressure changes. Line pressure monitoring
during infusions in gel has been scarcely reported on in
the literature. Infusion pressure monitoring may be useful to
detect catheter endport occlusions identified as initial infusion
pressure spikes (figure 2(B)). We have found that pressure
monitoring may be helpful in identifying possible problems
with the equipment such as leaks, air bubbles and improper
pump setting. Figure 2 highlights a possible protocol deviation
in VT catheter operation. The stylet may not have been
retracted prior to the beginning of the second infusion and as
a result a higher pressure recording was observed. In another
experiment (data not shown), abnormal pressure readings were
observed and it was later discovered that the ferrule used at the
female luer end had been damaged and was leaking. Pressure
measurements are also reported to be quite sensitive to air
pockets within the line tubing and may cause performance
deviations.
Length and inner diameters differ between the two
catheters used in our experiments. As a result, pressure
readings differ depending on the length of the catheter although
infusion performance appears to be independent of the catheter
length. Nevertheless, caution is advised when attempting to
compare infusion pressures across catheter types and lengths.
A more useful method to monitor infusion pressures may be
to obtain a baseline pressure for the characterization of a given
infusion system and monitor the resulting offset for initial
infusion pressure spikes or unexpected changes.
A trend was observed with respect to the initial infusion
pressure spike analysis when the categories are divided into
two subgroups. An example of a pressure spike reading can be
10
K Sillay et al
seen in figure 2(B). It appears that if an initial pressure spike
is observed, the chance of also observing backflow is 15 of
31. Conversely, however, if no initial pressure spike is noted,
the chance of observing backflow is only 5 of 17. This may
provide a measure of confidence that no backflow has occurred
during an infusion.
5. Conclusions
We report for the first time computer-controlled ramped
infusions benchmarking the VT and SF catheters in the first
phase of work designed to inform current and future human
clinical trials involving convection-enhanced delivery. Further
study is needed with ex vivo and in vivo studies to optimize
infusion parameters.
Results from the current study show comparable
performance between the VT catheter described and the
recently FDA approved SF catheter in a gel model. Simulations
of the infusion rate versus backflow would indicate a
relationship between the infusion rate at 1 versus 5 µL min−1.
The current work was unable to document a statistically
significant relationship. Further study regarding ramped versus
non-ramped protocols is indicated including the method and
duration of the ramp to maximum infusion.
With the emergence of real-time MRI monitoring of
infusions, further work is indicated in determining optimum
methods of targeting, monitoring infusions and providing
protocol-driven countermeasures to address such intra-
infusion findings. We foresee that patient operative time,
morbidity, cost and the surgical team’s learning curve can
be significantly reduced with these investigations. For the
first time, real-time imaging may confirm the accuracy of
targeting and infusion monitoring simultaneously, allowing
for a physician-directed go/no-go decision at several pause
points during the infusion procedure.
Acknowledgments
The authors acknowledge financial support from the Kinetics
Foundation. In addition, the authors would like to thank Martin
Brady and Raghu Raghavan, Therataxis, LLC, JHU Eastern
complex for the useful discussion.
Appendix A. Agarose gel protocol
Making an agarose gel:
1. Add 100 mL of 10XTBE Buffer to 900 mL of deionized
water to make 1 l of solution (resulting in a 1XTBE
solution).
2. Transfer the solution to a 1000 ml beaker.
3. Add 2 g of agarose powder to the solution to make a 0.2%
concentration.
4. Put the beaker in a microwave and heat until the boiling
point or until the solution is completely clear (this should
take approximately 8 min). Heat the solution in the
microwave in 3 min incrementsand stir. Remove using
heat gloves.
J. Neural Eng. 9 (2012) 026009
5. Measure the temperature of the solution until it reaches
85 ◦ C and then pour the solution into appropriate
containers. Leave 1 inch of space from the top of the
container.
6. Allow the solution to cool and solidify (for approximately
1–2 h) with the cap on.
7. Once the solution turns into a gel, add 1/2 an inch of water
to the surface of the gel.
8. Store the gel in a fridge/freezer (4 ◦ C) to help it cool and
to temporarily preserve it. It is best if the gel is used within
24–48 h after being made, but it can be kept for a week if
it is stored in the fridge/freezer.
Appendix B. Infusion protocol
Prepare loading lines
1. Cut the end of the tubing, leaving a square-cut face.
2. Slide the flangeless nut over the tubing, with the nut
threads facing the tubing end being connected.
3. Slip the flangeless ferrule over the tubing, with the tapered
portion of the ferrule facing toward the nut.
4. Insert the tubing with the ferrule in place into the receiving
part, and while holding the tubing down firmly into the
port, tighten the nut using either the male or female luer
finger right.
5. Connect the controller to the pump using a thick gray
cable and the controller to the computer with a phone
cable.
6. Connect the Male Luer AsyTefzel to the catheter.
7. Use a 60 ml syringe to inject distilled water into the tubing
and through the entire catheter system. (You may need to
lift the catheter from the distilled water it is being stored
in to see the water exiting the catheter.)
Setup
1. Prepare the infusion syringe by loading the infusate
carefully and ensuring all air is cleared from the syringe
and the infusate meniscus accumulates at the syringe luer
connection.
2. Attach the syringe and secure it to the pump.
3. Connect the pressure monitor attachment to the syringe.
Turn on the pump and run at 200 µl min−1 until the
dye reaches the luer connection of the pressure monitor
attachment. Make sure that the internal pressure sensor is
enveloped with dye.
4. Prepare cords by filling the rim of the nut with water
to prevent formation of air bubbles. Connect the female
luer to the pressure monitor attachment and connect the
pressure monitor attachment to the computer. Attach the
male luer to the tubing coming from the catheter.
5. Flush the tubes with the bromophenol blue dye at
200 µl min−1 until it is visible in the distilled water
the catheter is being stored in. Remove the catheters
from distilled water once pressure reading returns to the
baseline. (Make sure that the dye reaches the tip of the
11
K Sillay et al
catheter and is not pushed up by any washout of distilled
water, which can delay the infusion.) Use this as an
opportunity to check the connections by searching for
air bubbles and leaks.
6. Replace the lids on top of the gel with those that are
assigned to each catheter (VT on the left and SV on the
right).
7. Set up the camcorder (align the camera to include
appropriate data recordings on the screen such as infusion
pressure, pump rate and elapsed time while avoiding
screen blockage from the catheters).
8. Prepare the pressure monitor computer:
a. Enter MRI time
b. Enter syringe diameter
c. Zero volumes
d. Zero the pressure
e. Set program script.
Infusion
1. Start camera recording. Start pressure recording. State the
date and type of experiment.
2. Insert the VT catheter first. Just before inserting the
catheter completely into the gel zero the pressure while
the tip is immersed in the distilled water on the surface.
Continue by guiding the catheter by hand. Avoid twisting.
Continue to lower the catheter 10 mm at a consistent pace
using the remote introducer. Do this all with the pump
turned off. (No dye should travel through the catheter.)
3. Insert the SV catheter next. Just before inserting the
catheter completely into the gel zero the pressure while
the tip is immersed in the distilled water on the surface.
4. Retract the stylet for the VT catheter.
5. Start both pumps to begin the infusion.
6. Immediately continue inserting the catheter into the gel
by hand until both catheters are at the same depth in the
gel (measure this before hand and use the stopper as a
guide).
7. Take a picture of the gel at every change in the infusion
rate (approximately every 5 min).
8. Once the infusion reaches a final volume for both
catheters, halt the infusion, advance the VT catheter stylet,
and use the remote introducer to insert the VT catheter
15–20 mm further into the gel.
9. Repeat steps 4–8 until desired number of infusions is
completed.
10. Turn off the pump.
Post-infusion
1. Wait at least 10 min before removing the catheters from
the gel (try to do this as gently and slowly as possible.
Rotate the catheter slightly in order to loosen the catheter
from the gel to prevent suck back).
2. Take a picture of the gel 5, 10, 15, 20, 25 and 30 min after
infusion.
3. Take a final snapshot on the pressure recording program.
4. Move pressure files to the directory folder.
J. Neural Eng. 9 (2012) 026009 K Sillay et al

5. Check and record the final volume in the syringe after (8) UCSF 3.0 R—ramped protocol from 1.0 to 3.0 µl min−1 in
infusion. 0.5 µl steps every 5 min as described in the Richardson
6. Inspect the catheter and infusion line for occlusion/ et al (2011) paper.
leak/abnormality. (9) UCSF 5.0 R—ramped protocol from 1.0 to
5.0 µl min−1 in 1.0 µl steps every 5 min as described
in the Richardson et al (2011) paper.
Appendix C. Summary of infusion clouds (10) UW—University of Wisconsin, Madison.
(11) UW 1.0 C—continuous infusion at 1.0 µl min−1.
VT catheter SV catheter (12) Vd—volume of distribution; calculated volume of dye
ID Protocol ID Protocol diffusion in an infusion cloud at a given time point
modification of the programming script.
110413-1V UW 1.0 C 110413-1S UCSF 3.0 R (13) Vi—volume infused; computer-controlled volume of
110413-2V UW 1.0 C 110413-2S UCSF 3.0 R
infusate administered at a given time point.
110413-3V UW 1.0 C 110425-1S UCSF 3.0 R
110425-1V UW 1.0 C 110425-2S UCSF 3.0 R
110425-2V UW 1.0 C 110425-3S UCSF 3.0 R References
110425-3V UW 1.0 C 110426-1S UCSF 3.0 R
110426-1V UW 1.0 C 110426-2S UCSF 3.0 R Bankiewicz K S, Eberling J L, Kohutnicka M, Jagust W, Pivirotto P,
110426-2V UW 1.0 C 110426-3S UCSF 3.0 R Bringas J, Cunningham J, Budinger T F and Harvey-White J
110426-3V UW 1.0 C 110426-4S UCSF 3.0 R 2000 Convection-enhanced delivery of AAV vector in
110427-1V UW 1.0 C 110427-1S UCSF 3.0 R parkinsonian monkeys; in vivo detection of gene expression
110427-2V UW 1.0 C 110427-2S UCSF 3.0 R and restoration of dopaminergic function using pro-drug
110429-1V UW 1.0 C 110429-1S UCSF 3.0 R approach Exp. Neurol. 164 2–14
110429-2V UW 1.0 C 110429-2S UCSF 3.0 R Bobo R H, Laske D W, Akbasak A, Morrison P F, Dedrick R L
110429-3V UW 1.0 C 110429-3S UCSF 3.0 R and Oldfield E H 1994 Convection-enhanced delivery of
110429-4V UW 1.0 C 110429-4S UCSF 3.0 R macromolecules in the brain Proc. Natl Acad. Sci. USA
110511-1V UW 1.0 C 110511-1S UCSF 3.0 R 91 2076–80
110511-2V UW 1.0 C 110511-2S UCSF 3.0 R Chen Z J, Broaddus W C, Viswanathan R R, Raghavan R
110513-1V UCSF 3.0 R 110512-1S UCSF 3.0 R and Gillies G T 2002 Intraparenchymal drug delivery via
110519-1V UCSF 5.0 R 110512-2S UCSF 3.0 R positive-pressure infusion: experimental and modeling studies
110519-2V UCSF 5.0 R 110519-1S UCSF 5.0 R of poroelasticity in brain phantom gels IEEE Trans. Biomed.
110520-1V UCSF 5.0 R 110519-2S UCSF 5.0 R Eng. 49 85–96
110523-1V UCSF 5.0 R 110520-1S UCSF 5.0 R Chen Z J, Gillies G T, Broaddus W C, Prabhu S S, Fillmore H,
110524-1V UCSF 5.0 R 110523-1S UCSF 5.0 R Mitchell R M, Corwin F D and Fatouros P P 2004 A realistic
110524-2V UCSF 5.0 R 110524-1S UCSF 5.0 R brain tissue phantom for intraparenchymal infusion studies
110524-3V UCSF 5.0 R 110524-2S UCSF 5.0 R J. Neurosurg. 101 314–22
110525-1V UCSF 5.0 R 110526-3S UCSF 5.0 R Emborg M E et al 2010 Intraoperative intracerebral MRI-guided
110525-2V UCSF 5.0 R 110525-1S UCSF 5.0 R navigation for accurate targeting in nonhuman primates Cell
110525-3V UCSF 5.0 R 110525-2S UCSF 5.0 R Transplant 19 1587–97
110526-1V UCSF 5.0 R 110525-3S UCSF 5.0 R Morrison P F, Chen M Y, Chadwick R S, Lonser R R
110526-2V UCSF 5.0 R 110526-1S UCSF 5.0 R and Oldfield E H 1999 Focal delivery during direct infusion to
110526-3V UCSF 5.0 R 110526-2S UCSF 5.0 R brain: role of flow rate, catheter diameter, and tissue mechanics
110526-3S UCSF 5.0 R Am. J. Physiol. 277 R1218–29
Morrison P F, Laske D W, Bobo H, Oldfield E H and Dedrick R L
Summary of all infusions performed by the catheter and
1994 High-flow microinfusion: tissue penetration and
protocol used (n = 63).
pharmacodynamics Am. J. Physiol. 266 R292–305
Nicholson C and Tao L 1993 Hindered diffusion of high molecular
weight compounds in brain extracellular microenvironment
Appendix D. Definitions measured with integrative optical imaging Biophys. J.
65 2277–90
(1) AAV2-GDNF—adeno-associated virus serotype 2-glial-
Panse S J, Fillmore H L, Chen Z J, Gillies G T and Broaddus W C
derived neurotrophic factor. 2010 A novel coaxial tube catheter for central nervous system
(2) Backflow—the distance traveled by the infusate along infusions: performance characteristics in brain phantom gel
the catheter–gel interface measured from the tip of the J. Med. Eng. Technol. 35 408–14
catheter; this measurement was only made if the distance Raghavan R, Brady M L, Rodriguez-Ponce M I, Hartlep A,
Pedain C and Sampson J H 2006 Convection-enhanced
the dye traveled up the catheter–gel interface exceeded delivery of therapeutics for brain disease, and its optimization
the distance the dye traveled into it. Neurosurg. Focus 20 E12
(3) CED infusion cloud—individual masses of infusate Raghavan R, Mikaelian S, Brady M and Chen Z J 2010 Fluid
generated in agarose gel. infusions from catheters into elastic tissue: I. Azimuthally
symmetric backflow in homogeneous media Phys. Med. Biol.
(4) IIPS—initial infusion pressure spike. 55 281–304
(5) Infusion—injection of the solvent driven by the computer- Ratliff J K and Oldfield E H 2001 Convection-enhanced delivery in
controlled pressure gradient. intact and lesioned peripheral nerve J. Neurosurg.
(6) Ramped protocol—infusion rates set to change during the 95 1001–11
Richardson R M et al 2011 Interventional MRI-guided putaminal
infusion process at specific time points. delivery of AAV2-GDNF for a planned clinical trial in
(7) UCSF—University of California, San Francisco. Parkinson’s disease Mol. Ther. 19 1048–57

12
J. Neural Eng. 9 (2012) 026009
Rogawski M A 2009 Convection-enhanced delivery in the treatment
of epilepsy Neurotherapeutics 6 344–51
Sampson J H et al 2007 Intracerebral infusate distribution by
convection-enhanced delivery in humans with malignant
gliomas: descriptive effects of target anatomy and catheter
positioning Neurosurgery 60 ONS89–98 discussion
ONS-9
13
K Sillay et al
Schantz E J and Lauffer M A 1962 Diffusion measurements in agar
gel Biochemistry 1 658–63
White E, Bienemann A, Malone J, Megraw L, Bunnun C,
Wyatt M and Gill S 2011 An evaluation of the relationships
between catheter design and tissue mechanics in achieving
high-flow convection-enhanced delivery J. Neurosci. Methods
199 87–97