AJR 2004; 183:513-517
© American Roentgen Ray Society
Model to Quantify Lymph Node Enhancement on Indirect Sonographic Lymphography
Sang-Hee Choi1,2,
Yuko Kono1,
Jacqueline Corbeil1,
Olivier Lucidarme1,3 and
Robert F. Mattrey1
1 Department of Radiology, University of California, San Diego, 200 W Arbor Dr.,
San Diego, CA 92103.
Received September 25, 2003;
accepted after revision April 19, 2004.
Address correspondence to R. F. Mattrey.
2 Present address: Department of Radiology and Center for Imaging Science,
Sungkyunkwan University, Seoul, Korea.
3 Present address: Laboratoire d'Imagerie Parametrique, UMR 7623
CNRS-University Paris VI and Assistance Publique Hôpitaux de Paris,
Paris, France.
Abstract
OBJECTIVE. Our goal was to develop a reliable technique that has
minimal operator dependence for quantifying lymph node enhancement to test and
optimize new sonography contrast formulations.
MATERIALS AND METHODS. Twenty healthy rabbits were studied using
five agents, labeled A-G. Agents D and E were the same agent and agents F and
G were Imagent, studied blindly to test reproducibility. One milliliter of
contrast agent was injected into each hind footpad. A 13-MHz transducer was
fixed over the popliteal node, which was imaged at a 4.8-MHz central transmit
frequency using phase-inversion technology at 100% power and one frame per
second. Immediately after each injection, the footpad was massaged 12 times
for 30 sec each time and then imaged after each massage to assess the number
of times the node could be refilled from each injection. Lymph node video
intensity was measured, and the degree of enhancement was evaluated using
analysis of variance with the massage number and the agent used as independent
variables.
RESULTS. Lymph node enhancement was observed after the first massage
with all agents. Degree of enhancement was least with agents A and B,
intermediate with agents D and F, and greatest with agent C. Agent A was
effective after the first two massages, agent B after the first four, agent C
after all 12, agent D after the first eight, and agent F after the first nine.
Performance of agents D and F was similar to that of their duplicates, E and
G.
CONCLUSION. We established a reproducible technique to quantify
lymph node enhancement that can distinguish between different agents. The
differences in performance suggest that it is possible to optimize agent
formulation for indirect sonographic lymphography.
Introduction
Localization of lymph nodes in the drainage field of malignant lesions to
assess tumor spread has had a significant impact on treatment of breast cancer
patients because the degree of nodal involvement remains the most important
prognostic indicator in breast cancer
[1]. Because of significant
morbidity associated with axillary dissection, sentinel node resection, which
was popularized by Morton et al.
[2] for staging melanoma and
used for breast cancer by Giuliano et al.
[3], has become the preferred
technique to stage disease of the axilla. The difficulty of the procedure lies
in the localization and identification of the sentinel node. The sentinel
lymph node is the first lymph node in the lymphatic drainage path of the tumor
and is therefore the first node exposed to metastatic tumor cells. Although
sentinel node assessment has a less than 1% false-negative rate
[3,
4], sntinel nodes only detected
in 83% of cases [4]. Although
Giuliano et al. injected a water-soluble blue dye in and around the tumor to
turn the sentinel node blue and distinguish it from breast tissue
[3], the preferred technique is
the use of filtered technetium-99m sulfur colloid, imaging the axilla
preoperatively as a guide; the use of a pencil probe to locate the node
intraoperatively; and the administration of blue dye to allow nodal
visualization. Although radiolabeled colloids have a more delayed transit and
provide a skin-marking option, they are less than ideal. The fluid is
invisible intraoperatively; many nodes are enhanced; and, more important, the
proximity of the injection site to the nodes decreases the
target-to-background ratio, decreasing sentinel node specificity
[5].
We propose the use of microbubble-based sonographic contrast material as a
tool to locate the sentinel node preoperatively
[6]. Particles injected
subcutaneously enter the lymph vessel through gaps between lymphatic
endothelial cells or by transcellular endo- or exocytosis
[7]. On average, small
particles (10-40 nm) are more likely to enter than large ones. As particles
approach 1 µm, their uptake into the lymphatics is poor and they
must be carried away by phagocytes or reduced in size by local processes. In
fact, more than 95% of particles larger than 400 nm stay at the injection
site, whereas 74% of particles 10 times smaller (40 nm) are absorbed
[8]. Despite the preference for
using small particles, we hypothesized that, because sonography is sensitive
to microbubbles and because microbubbles are deformable and have a size
distribution that could contain 1-µm or smaller microbubbles,
lymph node enhancement would be observed. This hypothesis was shown to be true
using rabbits [6]. In our
feasibility study, we used a Food and Drug Administration (FDA)-approved IV
contrast agent designed for cardiac imaging. We noted extreme variability in
nodal enhancement and the ability to refill the nodes by remassaging the
injection site a variable number of times
[6]. We expect improvement in
performance with optimization of formulation; however, this improvement
requires the testing of agents in a reproducible and reliable model. The
purpose of this study was to develop a model with minimal operator dependence
and to assess its reproducibility by testing different agents and determining
whether the model can discriminate among their performances.
Materials and Methods
Animals
Twenty healthy female New Zealand White rabbits weighing 2.5-3 kg were used
in the study. They were anesthetized with 50 mg/kg of ketamine (Fort Dodge
Animal Health) and 8.8 mg/kg of xylazine given subcutaneously. Once anesthesia
was initiated, additional quarter doses were given as needed. After imaging,
animals were sacrificed with 150 mg/kg of pentobarbital.
The protocol was reviewed and received approval by our institutional animal
subjects committee in accordance with the policies of the United States
Department of Agriculture, Department of Health and Human Services, and
National Institutes of Health regarding the humane care and use of laboratory
animals.
Sonographic Contrast Media
Five agents were used that were made by IMCOR Pharmaceutical scientists,
the fifth being Imagent (AF0150), an FDA-approved IV contrast agent. Imagent
was manufactured in a good manufacturing practice facility to exact
specification for clinical use. With the exception of Imagent, the research
team was blinded to agent specifications until the experiment was completed.
Agent D was the identical formulation as Imagent except that it was
manufactured in the pilot plant under the same experimental conditions as
agents A, B, and C. Agent C had a modified emulsifier to minimize aggregation.
Agents A and B were similar to agents C and D except for the use of two
fluorocarbon compounds instead of only perfluorohexane. The four experimental
agents were labeled A, B, C, D, and E, with D and E being the same agent.
AF0150 was labeled agents F and G. Agent D and AF0150 were duplicated to
assess the reproducibility of the model and to blind the operator. All agents
consisted of heat-sterilized 200-mg powder composed of porous hollow
microspheres in a vial filled with a mixture of perfluorocarbon vapor and
nitrogen. All agents consisted of water-soluble structural agents,
surfactants, buffers, and salts. After reconstitution with 10 mL of sterile
water, gas microbubbles encapsulated by a thin lipid layer are formed that are
suspended in a solution buffered to a neutral pH that is isotonic with plasma.
The sterile water is injected through a double-lumen spike, and doses are
withdrawn through a 5-µm filter to trap any undissolved powder.
Imaging Protocol
All agents were administered as a 1-mL injection into the rabbit's footpad,
which was shaved to facilitate injection into the web. The rabbit's foot was
then placed between the rollers of a modified peristaltic pump
(Fig. 1) that massaged the
footpad in the direction of the popliteal fossa for 30 sec at a constant
rolling rate and pressure.
View larger version (109K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 1. —Photograph of experimental setup shows rabbit positioned in support
cradle. Injected leg was inserted into roller pump holder and massaged toward
popliteal fossa, and 13-MHz transducer was positioned over popliteal lymph
node and mechanically fixed in position during all 12 massages after each
injection.
|
|
The popliteal fossa was shaved for imaging, and, once the popliteal node
was located, the transducer was mechanically fixed over the node and
maintained at the same location for the entire experiment. A 13-MHz VFX
transducer from a Sonoline Elegra system (Siemens Ultrasound) equipped with
intermittent imaging capability and phase-inversion imaging was used for all
animals. The central transmit frequency was set at 4.8 MHz and was delivered
at 100% power with acquisition at one frame per second. With the depth of
field set at 1.5-2 cm, the focal zone was positioned at 0.5-1 cm from the
surface at the level of the popliteal node. With the exception of overall
gain, which was optimized for each animal at the beginning of each scanning
session, all imaging parameters were constant for all animals.
Agents A, B, and C were injected into eight footpads each (four rabbits)
and agents D, E, F, and G were injected into four footpads each (two rabbits).
The imaging protocol consisted of obtaining 10 images of the popliteal node
before the 30-sec massage to eliminate any residual bubbles from the prior
massage and four images immediately after massage. All 14 images were saved to
disk. The imaging protocol and massage were repeated 11 more times, resulting
in 12 massages after each injection for studying the differences among agents
over multiple massages.
Data Collection
The video intensity (gray levels 0-255) of the lymph node at baseline (10th
frame of premassage image series) and the video intensity on the first frame
acquired after each massage were measured off-line using Scion Image (Scion),
and the difference was calculated. The difference is considered to be the
degree of nodal enhancement produced by the massage. Nodes were grouped by
agent used and massage number, and the mean enhancement and standard error of
the mean were calculated.
Statistical Analysis
Degree of enhancement was evaluated for statistical significance using
analysis of variance with agent and massage number as independent variables
and the degree of enhancement as the dependent variable. Analysis was
performed on the PC version of JMP software (SAS Institute). Two analyses were
performed. The four legs injected with agents D, E, F, and G were compared for
statistical significance. The four legs injected with agent E were grouped
with the legs injected with agent D (agents D and E are the same agent) and
those of agent G were combined with agent F (agents F and G are the same
agent) resulting in eight legs for each of the five agents A, B, C, D, and F,
and the statistical analysis was repeated. Maximal enhancement between groups
was evaluated for statistical significance using an unpaired Student's
t test. When statistical significance was noted on the two-way
analysis of variance, a paired Student's t test was performed to
determine which data points were significantly different from each other and
from the baseline. Statistical significance was achieved if the p
value was 0.05 or less.
Results
All agents markedly enhanced the popliteal lymph node (Fig.
2A,
2B) on the first frame after
the first 30-sec massage. The degree of enhancement is shown in
Figure 3.
View larger version (84K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 2A. —Popliteal lymph node in rabbit. Sonograms of popliteal lymph node
obtained before (A) and after (B) second 30-sec massage after
administration of 1 mL of contrast material into right footpad. Note that
hypoechoic lymph node (arrowheads, A) seen before contrast
enhancement is markedly enhanced (arrow, B) after contrast
agent is administered.
|
|
View larger version (86K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 2B. —Popliteal lymph node in rabbit. Sonograms of popliteal lymph node
obtained before (A) and after (B) second 30-sec massage after
administration of 1 mL of contrast material into right footpad. Note that
hypoechoic lymph node (arrowheads, A) seen before contrast
enhancement is markedly enhanced (arrow, B) after contrast
agent is administered.
|
|
View larger version (12K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 3. —Graph shows baseline-adjusted lymph node enhancement expressed as
gray levels (0-255) observed on first frame after first massage after 1 mL of
contrast administration in footpad. Each agent was evaluated in eight legs.
SEM = standard error of mean.
|
|
Figure 4 shows the degree of
enhancement achieved on the first frame after each massage after a single
injection. The degree of enhancement was statistically dependent on the agent
used (p < 0.0001) and the massage number (p < 0.0001).
No interaction occurred between the independent variables, agent used, and
massage number (p > 0.997). Agents A and B were not statistically
different, nor were agents D and F. However, agent C was different from all
other agents (p < 0.0001), and the pair of agents D and F were
different from agents A and B (p < 0.001). The degree of
enhancement on the first frame after massage was dependent on massage number
for all agents. When assessed for statistically significant enhancement after
each massage, degree of enhancement was significant for agent A after the
first and second massages, agent B after the first four massages, agent C
after all 12 massages, agent D after massages 1-8 and agent F after massages
1-9.
View larger version (22K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 4. —Graph shows ability to extract additional contrast material from
injection site after subsequent massages to reenhance lymph node as function
of massage number. Effect of massage number on baseline-adjusted lymph node
enhancement in gray levels (0-255) was statistically significant for all
agents. SEM = standard error of mean.
|
|
When evaluating the reproducibility of the model by comparing the
performance of the same agents D and E and the same agents F and G using
blinded observers, the performance of agent D was the same as agent E
(p = 0.807) (Fig. 5A)
and that of agent F was the same as agent G (p = 0.693)
(Fig. 5B). The performance
relative to massage number was statistically significant for agents D and E
(p = 0.003) and agents F and G (p = 0.048).
View larger version (13K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 5A. —Graphs of lymph node enhancement. Graphs show comparison of
baseline-adjusted lymph node enhancement in gray levels (0-255) achieved with
same agents D and E (A) and F and G (B) as function of massage
number to assess reproducibility of model. Agent D (dotted line,
A) is statistically similar to agent E (solid line, A),
and agent F (dotted line, B) is statistically similar to agent
G (solid line, B). SEM = standard error of mean.
|
|
View larger version (14K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 5B. —Graphs of lymph node enhancement. Graphs show comparison of
baseline-adjusted lymph node enhancement in gray levels (0-255) achieved with
same agents D and E (A) and F and G (B) as function of massage
number to assess reproducibility of model. Agent D (dotted line,
A) is statistically similar to agent E (solid line, A),
and agent F (dotted line, B) is statistically similar to agent
G (solid line, B). SEM = standard error of mean.
|
|
Discussion
Lymphatic spread of malignancies is a major factor influencing the
prognosis of patients with cancer. Evaluation of regional and distant lymph
nodes provides critical information for treatment
[9-11].
Thirty to fifty percent of patients with metastatic cancer
[12,
13] and 80% of patients with
advanced stages of cancer have nodal involvement at initial diagnosis
[14]. Noninvasive imaging
techniques rely mainly on nodal shape and size to assess its malignant
potential. Because a significant number of metastatic deposits are found in
lymph nodes smaller than 5 mm and even more in lymph nodes smaller than 1 cm
[10,
13], specific contrast agents
that accumulate in nodes are needed. However, the false-positive and
false-negative rates of such an approach require clinical testing because
microscopic deposits will likely be missed and filling defects may not be
malignant.
Because the ultimate nodal assessment requires histologic analysis,
sentinel lymph node resection is now considered the standard approach for
axillary staging in patients with breast cancer. The long-term goal of our
effort is to convert sentinel node resection into a minimally invasive
procedure that can be performed under local anesthesia. To accomplish this
goal, high-performance reliable contrast media are needed that clearly direct
the operator to the sentinel node. Despite all the disadvantages of blue
water-soluble dye in locating the sentinel node intraoperatively, the dye does
fill the lymph channels, whose blue color guides the surgeon to the draining
node. We showed in our original report that the lymphatic duct is visible
after the interstitial injection of microbubbles
[6]. It is possible, therefore,
to inject the agent near the tumor and observe with sonography the filling of
the lymph ducts that can then be traced to the sentinel node. Once identified,
the node can be marked and removed with a local incision. The critical
requirement for the success of this approach is that the agent must be readily
absorbed into the lymphatics, and subsequent massages of the injection site
should refill the lymphatic duct and node because microbubbles are destroyed
by sonography. Because of the variability in massage, injection, and imaging,
it is difficult to reproducibly and consistently enhance the node to compare
agents as optimization of formulation is sought.
This study described an animal model that can be used to optimize
formulations. The model minimized the dependence of the technique on the
operator and rendered each massage identical to the preceding massages to more
predictably assess the refilling of the lymphatics from the injection pool.
The results of this study show that the model is reproducible. The model
yielded similar data when the same agent was given to different rabbits and
imaged and analyzed by a blinded operator. Agents Imagent and D performed
similarly in two sets of four legs because they were the same agent, although
one was manufactured for clinical and one for experimental use.
Although the model was reproducible, it was still dependent on technical
parameters. The injection of pressure-sensitive microbubbles into a tight
small space can destroy the agent. The placement of the leg between the
rollers with proper pressure can affect the quality of the massage between
rabbits. The placement of the transducer over the popliteal node may not be
optimal and may move between massages. Therefore, care is still required to
maintain consistency; however, the requirements are similar to any other
experimental setup.
Although the intention was to produce agents to test the reproducibility of
the model, this study provides some insight into potential optimization
schemes. When the aggregating or surface interaction potential (agent C) was
decreased, the agent provided greater enhancement and allowed the refilling of
the node over 12 times from the same injection. Agents A and B were made by
adding a second fluorocarbon to agents D and C, respectively, that decreased
performance.
In conclusion, we have shown that a reproducible technique to quantify
lymph node enhancement is possible. Different agents had different
performances suggesting that formulations can be optimized to maximize lymph
node filling.
References
- Fisher ER, Costantino J, Fisher B, Redmond C. Pathologic findings
from the National Surgical Adjuvant Breast Project (Protocol 4): discriminants
for 15-year survival—National Surgical Adjuvant Breast and Bowel Project
investigators. Cancer1993; 71[suppl 6]:2141
-2150[Medline]
- Morton DL, Wen D-R, Wong JH, et al. Technical details of
intraoperative lymphatic mapping for early stage melanoma. Arch
Surg 1992;127:392
-399[Abstract]
- Giuliano AE, Kirgan DM, Geunther JM, Morton DL. Lymphatic mapping
and sentinel lymphadenectomy for breast cancer. Ann
Surg 1994;220:391
-401[Medline]
- Stewart KC, Lyster DM. Interstitial lymphoscintigraphy for
lymphatic mapping in surgical practice and research. J Invest
Surg 1997;10:249
-262[Medline]
- Kapteijn BA, Nieweg OE, Liem I, et al. Localizing the sentinel node
in cutaneous melanoma: gamma probe detection versus blue dye. Ann
Surg Oncol 1997;4:156
-160[Abstract]
- Mattrey RF, Kono Y, Baker K, Peterson T. Sentinel lymph node
imaging with microbubble ultrasound contrast. Acad
Radiol 2002;9[suppl 1]:S231
-S235
- Ikomi F, Hanna GK, Schmid-Schonbein GW. Mechanism of colloidal
particle uptake into the lymphatic system: basic study with percutaneous
lymphography. Radiology1995; 196:107
-113[Abstract/Free Full Text]
- Oussoren C, Zuidema J, Crommelin DJ, Storm G. Lymphatic uptake and
biodistribution of liposomes after subcutaneous injection. II. Influence of
liposomal size, lipid composition and lipid dose. Biochim Biophys
Acta 1997;1328:261
-272[Medline]
- Misselwitz B, Platzek J, Radüchel B, Oellinger JJ, Weinmann
H-J. Gadofluorine 8: initial experience with a new contrast medium for
interstitial MR lymphography. MAGMA1999; 8:190
-195
- Weissleder R, Elizondo G, Josephson L, et al. Experimental lymph
node metastases: enhanced detection with MR lymphography.
Radiology1989; 171:835
-839[Abstract/Free Full Text]
- Berek J, Hacker N, Fu Y, et al. Adenocarcinoma of the uterine
cervix: histologic variables associated with lymph node metastasis and
survival. Obstet Gynecol1985; 65:46
-52[Abstract/Free Full Text]
- Chen SS, Lee L. Indices of paraaortic and pelvic lymph node
metastases in epithelial carcinoma of the ovary. Gynecol
Oncol 1983;16:95
-100[Medline]
- Herrera-Ornales L, Justiniano J, Castillo N, Petrelli NJ, Strule
JP, Mittleman A. Metastases in small lymph nodes from colon cancer.
Arch Surg1987; 122:1253
-1256[Abstract]
- Lee JKT, Stanley RJ, Sagel SS, McClennan BL. Accuracy of CT in
detecting intraabdominal and pelvic lymph node metastases from pelvic cancers.
AJR 1978;131:675
-679[Abstract]
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
Top
Abstract
Introduction
