Eur J Nucl Med Mol Imaging (2008) 35:9991007

DOI 10.1007/s00259-007-0664-2

ORIGINAL ARTICLE

Dynamic bioluminescence imaging for quantitative tumour

burden assessment using IV or IP administration
of D-luciferin: effect on intensity, time kinetics
and repeatability of photon emission
Marleen Keyaerts & Jacob Verschueren & Tomas J. Bos &
Lea Olive Tchouate-Gainkam & Cindy Peleman &
Karine Breckpot & Chris Vanhove & Vicky Caveliers &
Axel Bossuyt & Tony Lahoutte

Received: 13 August 2007 / Accepted: 18 November 2007 / Published online: 4 January 2008
# Springer-Verlag 2007

Abstract
Introduction In vivo bioluminescence imaging (BLI) is a
promising technique for non-invasive tumour imaging. Dluciferin can be administrated intraperitonealy or intraveMarleen Keyaerts is a Ph. D. fellow of the Research Foundation
Flanders (Belgium; FWO).
M. Keyaerts : L. O. Tchouate-Gainkam : C. Peleman :
C. Vanhove : V. Caveliers : A. Bossuyt : T. Lahoutte
In Vivo Cellular and Molecular Imaging (ICMI) Laboratory,
Vrije Universiteit Brussel (VUB),
Brussels, Belgium
M. Keyaerts (*) : C. Vanhove : V. Caveliers :
A. Bossuyt : T. Lahoutte
Department of Nuclear Medicine,
University Hospital Brussels (UZ-Brussel),
Laarbeeklaan 101,
1090 Brussels, Belgium
e-mail: Marleen.Keyaerts@vub.ac.be
J. Verschueren
Bio-Imaging lab, Department of Biomedical Sciences, University
of Antwerp,
Antwerp, Belgium
T. J. Bos
Department of Haematology and Immunology,
Vrije Universiteit Brussel (VUB),
Brussels, Belgium
K. Breckpot
Laboratory of Molecular and Cellular Therapy,
Department of Physiology and Immunology,
Vrije Universiteit Brussel (VUB),
Brussels, Belgium

nously. This will influence its availability and, therefore,

the bioluminescent signal. The aim of this study is to
compare the repeatability of BLI measurement after IV
versus IP administration of D-luciferin and assess the
correlation between photon emission and histological cell
count both in vitro and in vivo.
Materials and methods Fluc-positive R1M cells were
subcutaneously inoculated in nu/nu mice. Dynamic BLI
was performed after IV or IP administration of D-luciferin.
Maximal photon emission (PEmax) was calculated. For
repeatability assessment, every acquisition was repeated
after 4 h and analysed using BlandAltman method. A
second group of animals was serially imaged, alternating IV
and IP administration up to 21 days. When mice were
killed, PEmax after IV administration was correlated with
histological cell number.
Results The coefficients of repeatability were 80.2% (IV)
versus 95.0% (IP). Time-to-peak is shorter, and its variance
lower for IV (p<0.0001). PEmax was 5.6 times higher for
IV. A trend was observed towards lower photon emission
per cell in larger tumours.
Conclusion IV administration offers better repeatability and
better sensitivity when compared to IP. In larger tumours,
multiple factors may contribute to underestimation of
tumour burden. It might, therefore, be beneficial to test
novel therapeutics on small tumours to enable an accurate
evaluation of tumour burden.
Keywords Tumour imaging . Small animal imaging .
Molecular imaging . Bioluminescence imaging .
Animal model

1000

Introduction
In the development of effective anti-tumour therapies, an
important role is reserved for testing on animal models. For
this purpose, a large number of subcutaneous but also
orthotopic tumour models are available in mice. The measurement of the response of the tumour to the therapy is the
key parameter that needs to be assessed. Serial monitoring of
tumour growth in vivo is mostly done by volumetric calliper
measurements [1]. However, this method is fairly inaccurate
because it is operator dependent, difficult to perform in the
early stages of tumour development, and impossible for
orthotopic tumours. Another option is to make histological
sections and count tumour cells. A limitation of this approach is that it is time-consuming and does not allow intraindividual assessment of response to treatment. Non-invasive
methods based on anatomical measurements of tumour tissue
(echography, CT or MRI) are used, but the relatively low
contrast between the tumour and the surrounding tissues
makes it difficult to accurately delineate the tumour.
Moreover, these measurements also include other tissues
within the tumour like blood vessels and necrosis. Metabolic
and reporter gene imaging using radionuclide tracers
overcome this drawback, but they demand the use of radioactive tracers and dedicated infrastructure. In vivo bioluminescence imaging (BLI) is a promising technique for
non-invasive high throughput tumour burden imaging [25].
It uses luciferase reporter genes that are naturally found in
arthropods and marine organisms, like fireflies and squids.
These enzymes convert a substrate (e.g. D-luciferin, coelenterazine) to an oxydated state, resulting in photon emission
in the visual spectrum [6, 7]. The administration route of the
substrate can be intraperitoneal or intravenous. This will
influence the availability of the substrate at the tumour site
and, as a consequence, the intensity and the time course of
the bioluminescent signal. IP administration requires resorption from the peritoneal cavity to the bloodstream that can
vary between and within animals. Therefore, we hypothesise
that quantification after IV administration is more accurate.
The aim of this study is to compare the repeatability of
the measurement of photon emission after IV versus IP
administration of D-luciferin and assess the correlation
between photon emission and histological cell count both
in vitro and in vivo.

Materials and methods

Construction and generation of luciferase-encoding
lentiviral particles.
The commercial vector pGL4.10 (Promega, Madison, WI,
USA) encoding firefly luciferase (luc2) codon optimised for

Eur J Nucl Med Mol Imaging (2008) 35:9991007

more efficient expression in mammalian cells was used.

The gene was excised by BglIIXbaI digestion and inserted
into the transfer plasmid pHRtripCMV-IRES-tNGFR-SIN
[8] to yield pHRtripCMV-luc2-IRES-tNGFR-SIN. The
Fluc gene is driven by a cytomegalovirus (CMV) promoter
and followed by the truncated nerve growth factor (tNGFR)
gene, separated by an internal ribosomal entry site (IRES).
Luciferase-encoding lentiviral vector particles were produced in 293T cells by transient cotransfection of the
transfer (pHRtripCMV-luc2-IRES-tNGFR-SIN), envelope
(pMD.G) and packaging plasmid (pCMVR8.9), previously described in [8]. The vector stock was collected 48
and 72 h posttransfection, and concentrated by ultracentrifugation as described in [9]. The viral titer was determined
by infection of 293T cells with serial dilutions of the vector
stock. Seventy-two hours after infection, the number of
luciferase-positive cells was determined by florescenceactivated cell sorting analysis.
Generation of a luciferase-positive R1M
rhabdomyosarcoma cell line
Fresh R1M rhabdomyosarcoma cells were transduced at a
multiplicity of infection of 30. Post-transduction, cells were
surface stained by standard techniques. Briefly, cells were
incubated with an in-house-made anti-human tNGFR
antibody (clone HB8737) for 30 min at 4C, washed,
incubated with a streptavidinPE-labelled antibody (BD
Pharmingen, Erembodegem-Aalst, Belgium) for 30 min at
4C and washed. Flow cytometry was used to evaluate the
transduction efficiency (percentage of tNGFR positive
cells) and viability of the R1M cells. The transduction
efficiency after lentiviral transduction was evaluated up to
32 days after transduction. We refer to the luciferaseexpressing cell line as R1M-Fluc and to the non-transduced
cells as R1M wild type (R1M-wt).
R1M cells were grown in minimal essential medium
with 10% fetal bovine serum, 1% non-essential amino
acids, 100 U/ml penicillin, 100 g/ml streptomycin and
0.13 g/ml fungizone (all from Invitrogen, Paisley, UK).
In vitro bioluminescence measurements
R1M-Fluc cells were plated in 96-well plates in triplicate in
a dilution series ranging from 1104 to 3102 cells. Four
hours after plating, D-luciferin (Xenogen, Alameda, CA)
was added to reach a final concentration of 0.15 mg/ml.
BLI of cells was performed in parallel on a GloMax 96
Luminometer (Promega, Madison, WI, USA), as well as in
a photo imager camera (Biospace, France) that allows listmode acquisition. For luminometer measurements, photon
emission was measured before (background) and 45 s after
automatic injection of D-luciferin during 10 s. Background

Eur J Nucl Med Mol Imaging (2008) 35:9991007

1001

was subtracted from the measured photon emission, and the

results are expressed in relative light units (RLU). For
camera measurements, D-luciferin was added using a
multichannel pipette, and time was measured between Dluciferin injection and start of the dynamic acquisition.
Identical circular regions of interest (ROI) were drawn over
the wells, and their time profiles using 1 s intervals was
analysed. The photon emission of R1M-wt-containing wells
did not rise above background and was used for background subtraction. The integrated photon emission over
10 s at 45 s after substrate administration was calculated to
compare the results from the charge-coupled device (CCD)
camera and the luminometer during the same time interval.
Additionally, the maximum photon emission (PEmax) of the
dynamic profile, up to 5 min after addition of D-luciferin,
was derived using the 95th percentile. From this value, the
mean PEmax per cell was calculated.

on gray-scale photographic images of the mice, with the

most intense detected luciferase signal shown as red and the
weakest signal shown as blue.
For image analysis, an elliptical ROI was drawn over the
tumour location. The surface area of the ROI was kept
constant. A time activity curve for every acquisition was
obtained by analysing images in 5-s intervals using photovision+ software (Biospace). For the calculation of the
maximal photon emission (PEmax) from the tumour, the
95th percentile of 5-s interval counts over 40 min after
substrate administration was used. This was expressed as
photons/s/steredian using the conversion factor (1 cpm=28
photons/s/steredian) provided by the camera manufacturer
(Biospace) and is referred to as PEmax. For the calculation
of the time-to-peak, the same acquisition data was analysed
using 1-min intervals, and the time point containing the
highest photon emission was defined as time-to-peak.

Animal model

Serial BLI of tumour burden

Male athymic nude (nu/nu) mice were purchased from

Harlan (The Netherlands) and were between 5 and 11 weeks
old before initiation of experiments. The study protocol was
approved by the Ethical Committee for Animal Studies of
our institution, and the National Institutes of Health
principles of laboratory animal care (NIH publication 86
23, revised 1995) were followed.
Fresh R1M-Fluc cells were collected with trypsin
ethylenediaminetetraacetic acid 0.05% (Invitrogen) and
resuspended in PBS. Under 2.5% isoflurane, mice were
injected subcutaneously into the flank with 100 l of cell
suspension containing 1105 cells using a 22-gauge needle.
Tumours were allowed to grow up to 21 days.

Six mice of the animal group described above were further

analysed to assess serial BLI of tumour burden after IV and
IP administration of D-luciferin. Animals were imaged
starting three (IP) or four (IV) days after tumour inoculation. For each day of tumour development, the average
SEM PEmax after IV or IP was calculated. Data of days 20
and 21 were not included in the analysis because less then
three measurements were available to calculate the average.
An exponential fit was performed on the data to derive the
tumour doubling time for both IV and IP.

Dynamic BLI after IV or IP administration

Twelve mice were imaged between days 3 and 21 post
tumour inoculation. Mice were anaesthetised with a mixture
of oxygen/isoflurane using an inhalation anesthesia system
(VetTech solutions, UK), 5% isoflurane for induction and
2.5% isoflurane for maintenance. D-luciferin was injected at
30 mg/kg mouse body weight either IP or IV via the tail
vein using a stock solution at 6 mg/ml. IV and IP
administration was alternated on subsequent days to assure
a cross-over study design (n=63 for IV, n=61 for IP
administration). Immediately after D-luciferin administration, mice were imaged using the photo imager (Biospace).
The photon emission was measured dynamically (list-mode
acquisition) using the large field-of-view setting and
registered using the photon counting technology (Biospace)
during 40 min. At the end of each acquisition, a
photographic image was obtained. Bioluminescent pseudocolor images displayed in the text are shown superimposed

Repeatability of bioluminescence measurements

A light phantom containing 16 standardised light sources
(Imager testing device TD216, Biostep, Germany) was used
to evaluate the repeatability of the photo imager camera
(Biospace) and repeated twice. Additionally, an absorptive
neutral-density filter with a nominal density at 546 nm of 3.0
(03 FNA 026, Melles Griot, CA, USA) was used to obtain a
1,000-fold reduction in the light from the weakest sources of
the phantom to obtain photon emissions that are in the range
of the lowest values that are generally measured in the in
vivo experiments. Different light sources were imaged twice
during 2 min with a 4-h interval using list-mode acquisition.
Image analysis was performed by drawing identical ROIs
over the different standardised light sources and by using 5-s
intervals. The mean photon emission was calculated and
expressed as photons/second/steredian (ph/s/sr).
A separate group of seven mice was imaged twice on the
same day with a 4-h interval using either IP or IV
administration at days 811 and 1315 after inoculation.
The acquisitions were performed as described above. Before
the second acquisition on the same day, residual biolumines-

1002

Eur J Nucl Med Mol Imaging (2008) 35:9991007

cence was measured in each mouse. Both IV and IP was

performed within the same mouse on different days. PEmax
was calculated for each acquisition, as was described above.
Dynamic versus static BLI
Each dynamic list mode acquisition of the groups dynamic
BLI after IV or IP administration and repeatability of
bioluminescence measurements was reevaluated retrospectively to simulate a static imaging protocol. Photon emission
during 1 min was calculated at the average time-to-peak for
IV and IP, which was obtained as described above. In the
group of dynamic BLI after IV or IP administration, static
photon emission was expressed as percent of the dynamic
PEmax. In the group repeatability of bioluminescence
measurements, analysis was performed using identical
statistical methods for static and dynamic imaging results.
Correlation of in vivo BLI and ex vivo tumour cell count
Quantitative histological analysis was performed on excised
tumours of a subgroup of animals (n=5) at days 1416 after
inoculation to allow comparison with in vivo PEmax after
IV administration of D-luciferin. Before dissection, mice
were imaged after IV injection of D-luciferin at 30 mg/kg
and 6 mg/ml. Dissection samples were fixated in paraformaldehyde (4%) and stored in 70% ethanol. After paraffin
embedding, the tissue samples were sectioned into 5-m
slices. Of every 30 sections, three subsequent sections were
mounted on slides and stained with haematoxylin and
eosin, and 27 sections were discarded. Analysis was
performed using the Cavalieri method as previously
prescribed in [10] and using a computerised stereological
system (CAST-grid, Olympus, Albertslund, Denmark). A
grid of crosses was superimposed onto the tissue samples.
By counting all the crosses touching the area of interest in
one section every 150 m and by using the following
formula, the volume can be calculated:
m
a X
V^ T


Pi
p i1

with Pi as the number of points landing within the object on

the ith section, T the distance between analysed sections
and a/p the area associated with each point. To determine
the amount of cells per volume, the same sections were
used. A physical disector was made with two subsequent
sections. A counting frame and meander sampling were
used to estimate the numerical density of cells using the
following formula:
^
N
v

P 
1
Q

P
a=f
h
P

^ as the estimate of numerical density, a/f the area of

with N
v
the frame, h the disector height, Q the total amount of
cells counted and P the sum of frame-associated points
hitting the reference volume. An estimate of the total
number of cells in a tumour was obtained by multiplying
^ . The
the volume V^ and the amount of cells per volume N
v
in vivo photon emission per cell was calculated by dividing
the PEmax by the estimated total tumour cell number.
Statistical analysis
Results are expressed as mean standard error of the mean
(SEM). Students t test was used to demonstrate significant
differences between groups. F test was used to assess
differences in variance.
To assess the repeatability of measurements, a Bland
Altman analysis was performed on the repeated measurements. Differences were expressed as percent of the average
of both measurements e1 and e2. The coefficient of
repeatability was calculated using the following formula:

CR 1:96 

v
2
u 
u e2 e1
 100
t@
e

n1

with CR as the coefficient of repeatability (in percent), e1

the value of the first measurement, e2 the value of the
second measurement, the average of e1 and e2, and n the
number of measurements.

Results
Correlation between cell number and photon
emission in vitro
The transduction efficiency of R1M-Fluc cells was evaluated up to 32 days after transduction, revealing only a
minor decrease in tNGFR positive cells (99.5% on day 3,
92.9% on day 32) without hampering their viability or in
vitro growth. Dilution series of R1M-Fluc cells were
analysed in triplicate during 10 s (45 s after administration)
by luminometer and CCD camera. Both show an excellent
correlation between photon emission and cell number
(Fig. 1). PEmax per cell was 4.480.33 ph/s/sr.
Dynamic BLI after IV or IP administration
Maximal photon emission after IV was on average 5.62.7
times higher compared to PEmax after IP. Figure 2 shows
bioluminescent images within the same mouse after IV and
IP administration of D-luciferin on subsequent days. The

Eur J Nucl Med Mol Imaging (2008) 35:9991007

1003

(between 41.5 and 102.5%) is wider for IP compared to IV.

The coefficient of repeatability for repeated IP measurements
is 95.0%, which is also higher as compared to IV.
Dynamic versus static BLI

Fig. 1 In vitro photon emission in function of cell number. Both

luminometer and CCD camera show an excellent correlation between
photon emission and cell number. CCD Charge-coupled device, RLU
relative light units, ph/s/sr photons/second/steredian

kinetic profile after IV administration shows an early peak

followed by a steep drop in photon emission (Fig. 3). The
profile after IP injection shows a gradual increase with a
relatively late peak followed by a slow decrease in photon
emission (Fig. 3). A stable plateau of photon emission was
not observed.
Time-to-peak was significantly shorter after IV administration: 5.70.6 min for IV and 24.71.2 min for IP (p<
0.0001, Fig. 4). The variance of time-to-peak was significantly lower for IV (p<0.0001, Fig. 4).
Serial BLI of tumour burden
Figure 5 shows the average increase in photon emission in
function of time after tumour inoculation within the same
animal group after IV and IP administration of the substrate.
For days with two or less successful acquisitions (days 20 and
21), no average was calculated. An exponential fit on the
curve results in a doubling time of 1.74 days for IV and
2.17 days for IP.

Each dynamic acquisition was reevaluated retrospectively

to simulate a static imaging protocol. Photon emission
during 1 min was calculated at the average time-to-peak
(5.7 min for IV, 24.7 min for IP). In the group of serial
tumour follow-up, static photon emission was expressed as
a percent of the dynamic PEmax, resulting in a significantly
higher value for IV (85.51.8%) than for IP (79.92.0%; p
< 0.05). In the repeatability study, the coefficient of
repeatability was recalculated using the same static simulation, resulting in an increase in coefficient of repeatability
from 80.2 to 81.2% for IV and from 95.0 to 101.7% for IP.
Correlation of in vivo BLI and ex vivo tumour cell count
To compare in vivo photon emission and ex vivo cell count,
mice were first imaged after IV administration of D-luciferin,
and tumours were subsequently dissected and analysed by
quantitative histology. Histological tumour volume and cell
density were used to calculate total tumour cell number. The
tumour volume ranged from 87 to 2,213 mm. No regions of
necrosis were identified during histological analysis. The
photon emission per cell in vivo is on average 0.11
0.02 ph/s/sr but is higher for lower cell numbers (Fig. 7).

Discussion
Bioluminescence is widely used and well validated for in
vitro measurement of adenosine triphosphate (ATP) [11, 12].

Repeatability of bioluminescence measurements

Repeated measurements of calibrated light sources, with
range of intensities that correspond to the range that is
measured in in vivo experiments, show excellent repeatability (Fig. 6a). The coefficient of repeatability is 3.6%.
This excludes that the variability of the BLI measurements
in animal studies is because of the detection system.
Repeated imaging for both IV and IP revealed no residual
photon emission above background after 4 h. The Bland
Altman plots for repeated measurements are shown in Fig. 6.
Repeatability of IV measurements shows a mean difference
of 29.0%. The corresponding 95% confidence interval ranges
between 25.4 and 83.3%. The coefficient of repeatability
for repeated IV measurements is 80.2%. Repeatability for IP
measurements shows a mean difference of 30.5%. This value
is comparable to the systematic error of repeated IV
measurements. The corresponding 95% confidence interval

Fig. 2 Bioluminescent overlay images of the same mouse bearing a

R1M-fluc positive tumour on the right flank after IV (a) and IP (b)
administration of D-luciferin on subsequent days. Both images show
the integrated light signal of 1 min at peak activity and are scaled to a
maximum of 28,000 ph/s/sr

1004

Eur J Nucl Med Mol Imaging (2008) 35:9991007

Fig. 3 Representative dynamic

profiles of the photon emission
after IV and IP administration of
D-luciferin. After IV injection,
an early peak is observed followed by a rapid decline. After
IP injection, a delayed peak is
measured with a more gradual
decrease thereafter

D-luciferin

is added in excess, which makes ATP the ratelimiting component. In vivo measurements are dependent
on multiple factors that alter photon emission [13]. Earlier
studies showed that the enzyme is not saturated at the
highest used dose of 150 mg/kg, as was theoretically
suggested [1416]. The intensity of the photon emission
changes with time, depending on the availability of the
substrate at the tumour site. This availability is directly
related to the pharmacokinetics of the administered substrate, and it can vary between animals [16]. Therefore, the
dynamic profile of photon emission is of importance for the
quantification of the signal. This time profile is not
efficiently sampled by a single static image. Measurement
of photon emission in function of time has been done using
consecutive static images [1619]. Camera systems using
list-mode acquisition enable the measurement of photon
emission with high-temporal resolution. This type of
dynamic BLI (D-BLI) is particularly of interest when IV
administration of D-luciferin is used because of its fast
pharmacokinetics.
After IP administration of D-luciferin, the first step in
biodistribution is the resorption from the peritoneal cavity
to the blood. This step can induce variability, as was noted
in previous studies [16]. Therefore, IV administration
should reduce any variability that might be linked to
differences in resorption between and within animals.
First, we looked at the differences in signal intensity and
signal kinetics after IV and IP administration of D-luciferin.
Second, we assessed repeatability of BLI measurements
using both administration routes. Third, we compared in
vitro to in vivo photon emission per cell after IV
administration of D-luciferin.
When comparing in vivo BLI after IV and IP administration of D-luciferin, IV shows shorter time-to-peak
intervals, less variability in time-to-peak intervals and
higher signal intensity than IP. The rapid peak of light
emission after IV injection is explained by the instant
delivery of the substrate to the tumour cells through the
bloodstream. The steep drop afterwards indicates the
elimination of the substrate from the blood. IP injection
has a less pronounced peak and gentler slope afterwards.
This is caused by the gradual release of the D-luciferin into

the bloodstream from the peritoneum. These differences in

kinetics for IV and IP administration were also reported by
Wang and El-Deiry [19]. Time-to-peak is more variable
with IP compared to IV injection. It is, therefore, more
likely to miss the peak after IP than after IV injection when
imaging. As a result, shorter acquisition procedures are
required when an IV injection is used, which limits the
anaesthesia time of the animals. Prolonged anaesthesia
might have detrimental effects on the organism and can
influence the enzyme reaction by changing the respiration
rate of the animal and, therefore, changing the oxygen
concentration at the tumour site. Shorter acquisition times
also allow for higher throughput. For some acquisitions, the
time-to-peak is extremely short after IP or long after IV
(Fig. 4), which can be explained by unnoticed extravasation
after IV injection, causing a late peak, and by accidental
intravascular injection during IP injection, resulting in a
rapid peak signal. The maximum intensity of photon
emission is about six times higher after IV injection
compared to IP injection and is explained by the bolus IV
administration of the substrate through which higher blood
and tissue concentrations are reached. It improves the
detectability of low cell numbers after IV injection and
reduces the amount of D-luciferin needed for equal
detectability of cells.

Fig. 4 Scatterplot of the time-to-peak values after IV and IP

administration of D-luciferin (days 321 post-tumour inoculation, 12
mice). Each data point represents the time of peak photon emission
after IV or IP administration. The time-to-peak is significantly shorter
and less variable for IV compared to IP

Eur J Nucl Med Mol Imaging (2008) 35:9991007

1005

Fig. 5 Photon emission after IV and IP administration of D-luciferin

in function of the time after inoculation (n=6). Results are expressed
as mean SEM using linear (a) and logarithmic (b) scale on the y-

axis. An exponential fit is performed to obtain the tumour doubling

time (IV, 1.74 days; IP, 2.17 days)

To assess the repeatability of BLI measurements, it

requires a measuring device that has a stable performance.
Quality control of the imaging devices is well standardised
for PET, SPECT, CT and MRI machines. In our study, we
performed a phantom study to verify the repeatability of the
photon emission measurements of the CCD camera
showing an excellent repeatability. These results indicate
that the relatively large variation in the results from the in
vivo experiments cannot be attributed to inherent instability
of the camera. The repeatability of in vivo BLI measurements is moderate for both IV and IP administration but is
better when light emission is measured after IV injection of
D-luciferin when compared to emission after IP injection.
The difference in repeatability between IV and IP might be
because of several causes. The pharmacokinetics of the
substrate in the body can be more variable after IP
administration. Resorption of D-luciferin from the peritoneal cavity to the blood can vary within the same animal and
between animals. This is not the case for IV administration.

A limitation of this experiment is the tumour growth during

the 4-h interval, which might vary between animals.
Serial tumour follow-up was performed to assess the
tumour model characteristics when BLI is used. Theoretically, tumours grow in an exponential way, and when therapy
is applied, tumours should deviate from this exponential
curve. When exponential fits are performed on the average
BLI signal over time, an excellent correlation is seen (r2
0.95 for both IV and IP). The doubling time of both
methods is similar and approximately 2 days.
The repeatability study shows a systematic error of about
30% over 4 h, which is higher than can be explained by the
exponential tumour growth with a doubling time of 2 days.
Other factors must be of influence, such as a decreased
binding to plasma proteins in case of a second administration or differences in enzyme turnover because of an earlier
contact with luciferin.
The general variability for both IV and IP can be caused
by multiple parameters. First of all, the rapid tumour

Fig. 6 Bland Altman plot of the repeated measurement of calibrated

light sources (a) and in vivo measurements of tumour burden using IV
administration (b) and IP administration (c) of D-luciferin. The
horizontal lines indicate mean (thick lines) and 95% confidence

intervals (thin line). Phantom repeatability is excellent. Repeatability

of IV and IP is moderate but better for IV compared to IP. COR
Coefficient of repeatability

1006

Fig. 7 Maximal photon emission in vivo in function of histological

cell count. The photon emission per cell in vivo is on average 0.11
0.02 ph/s/sr but is higher for lower cell numbers

growth can vary between animals, as well as within the

same animal. Secondly, differences in positioning of the
animal influence the amount and characteristics of overlying tissues, which change the penetration of light. Other
factors that could have an effect are minor differences in the
dose of the administered substrate, the percentage of
oxygen during anaesthesia, the anaesthesia and the temperature of the animal [20]. A standardised method of animal
preparation and imaging was applied to limit the effects
from these factors. Higher substrate dosage resulting in an
important increase in photon emission, as was shown by
Burgos et al. and Paroo et al., might help to reduce
variability, as here, only 30 mg/kg was used [16, 17]. This
is not unlimited, however, as doses above 150 mg/kg might
become toxic to the organism.
For both serial tumour follow-up and repeatability
assessment, the dynamically acquired data were compared
to simulated static imaging at average time of peak.
Although the fast kinetics after IV administration might
theoretically have an important effect on the intensity of the
measured signal if the peak is missed, this is not observed
in our values because of the low variation in time-to-peak
for IV measurements. On the contrary, the effect of single
static imaging is much larger after IP administration
because of the high variation in time-to-peak. This also
affects the repeatability, with only a minor decrease in
repeatability for static imaging after IV administration and a
more important decrease after IP administration of Dluciferin. A recent study by Baba et al. [21] compared
single static versus repeated static imaging after IP
administration, and they also saw a lower reproducibility
for single static imaging, although to a lesser degree than in
our study. In their study, the best single static imaging
results were obtained at 20 min post-injection [21].
For in vivo evaluation of an anti-cancer treatment using
the Fluc-R1M mouse model, IV administration of Dluciferin offers multiple advantages. The higher rate of

Eur J Nucl Med Mol Imaging (2008) 35:9991007

photon emission makes this method more sensitive than IP.

Lower variability in time-to-peak allows relatively short
acquisition procedures that reduce anaesthesia of the
animals. A 95 percentile of a dynamic acquisition of
10 min could be used as peak value and, hence, as a
measure of tumour viability. Static imaging of 1 min at
average peak photon emission only reduces the repeatability mildly. A drawback of IV injection is that it is more
difficult to perform which might result in a decrease of
successfully measured data points. IP injection is easier but
injections into the intestine or bladder are not noticed and
result in greater variability. To increase throughput, static
IP imaging can be performed but this decreases its
repeatability.
Correlation between cell number and bioluminescent
signal was studied both in vitro and in vivo. In vitro, intact
cells were imaged in dilution series. The excellent
correlation between photon emission and cell count
corresponds with previous findings on other cell cultures
[3, 22]. It indicates that, in stable conditions, photon
emission correlates linearly to the amount of cells. In vivo,
however, the correlation between cell number and bioluminescence is weaker. The limited number of data points does
not allow an assessment of the correlation, but there is a
trend towards lower photon emission per cell in larger
tumours. This could be explained by a lower substrate
concentration at the tumour site in vivo, with a relatively
poor penetration in larger tumours. Alternatively, this can
be because of attenuation of the light emission in vivo,
which is higher for larger tumours. Although in vitro results
only showed a slight decrease in reporter gene expression
over a 32-day interval, decrease in reporter gene expression
or selection of luciferase-negative tumour cells in vivo
cannot be excluded as a contributing factor. Underestimation
in large tumours has also been shown in previous studies that
compared bioluminescence to tumour weight or calliper
measurements, although also necrosis could have contributed to that discrepancy [23, 24]. As the quantification seems
to be more accurate for small tumours, it might be
beneficial to start early with experimental tumour therapies
to enable an accurate evaluation of tumour response.

Conclusion
In this Fluc-positive rhabdomyosarcoma mouse model, IV
administration of D-luciferin offers better repeatability,
better sensitivity and shorter imaging procedures when
compared to IP administration. In larger tumours, multiple
factors may contribute to an underestimation of tumour
burden when BLI is used. It might, therefore, be beneficial
to test novel therapeutics on small tumours to enable an
accurate evaluation of residual tumour burden.

Eur J Nucl Med Mol Imaging (2008) 35:9991007

Acknowledgement Marleen Keyaerts is a Ph. D. fellow of the
Research FoundationFlanders (Belgium; FWO). Tony Lahoutte is a
Senior Clinical Investigator of the Research FoundationFlanders
(Belgium; FWO). Karine Breckpot is a postdoctoral fellow of the
Research FoundationFlanders (Belgium; FWO). The research at
ICMI is funded by the Interuniversity Attraction Poles Programme
Belgian StateBelgian Science Policy.

References
1. Kim SA, Kim YC, Kim SW, et al. Antitumor activity of novel
indirubin derivatives in rat tumor model. Clin Cancer Res 2007;13
(1):2539.
2. Rehemtulla A, Stegman LD, Cardozo SJ, et al. Rapid and
quantitative assessment of cancer treatment response using in
vivo bioluminescence imaging. Neoplasia 2000;2(6):4915.
3. Jenkins DE, Oei Y, Hornig YS, et al. Bioluminescent imaging
(BLI) to improve and refine traditional murine models of
tumor growth and metastasis. Clin Exp Metastasis 2003;20
(8):73344.
4. Contag CH, Jenkins D, Contag PR, Negrin RS. Use of reporter
genes for optical measurements of neoplastic disease in vivo.
Neoplasia 2000;2(12):4152.
5. Contag PR, Olomu IN, Stevenson DK, Contag CH. Bioluminescent indicators in living mammals. Nat Med 1998;4(2):2457.
6. Hastings JW. Chemistries and colors of bioluminescent reactions:
a review. Gene 1996;173(1 Spec):511.
7. Wood KV, Lam YA, McElroy WD. Introduction to beetle
luciferases and their applications. J Biolumin Chemilumin
1989;4(1):289301.
8. Breckpot K, Dullaers M, Bonehill A, et al. Lentivirally transduced
dendritic cells as a tool for cancer immunotherapy. J Gene Med
2003;5(8):6547.
9. Breckpot K, Heirman C, De Greef C, van der Bruggen P,
Thielemans K. Identification of new antigenic peptide presented
by HLA-Cw7 and encoded by several MAGE genes using dendritic
cells transduced with lentiviruses. J Immunol 2004;172(4):22327.
10. De Jonge F, Van Nassauw L, Van Meir F, Miller HR, Van Marck
E, Timmermans JP. Temporal distribution of distinct mast cell
phenotypes during intestinal schistosomiasis in mice. Parasite
Immunol 2002;24(5):22531.

1007
11. Frundzhyan V, Ugarova N. Bioluminescent assay of total bacterial
contamination of drinking water. Luminescence 2007;22(3):2414.
12. Higashi T, Isomoto A, Tyuma I, Kakishita E, Uomoto M, Nagai K.
Quantitative and continuous analysis of ATP release from blood
platelets with firefly luciferase luminescence. Thromb Haemost
1985;53(1):659.
13. Levin CS. Primer on molecular imaging technology. Eur J Nucl
Med Mol Imaging 2005;32(Suppl 2):S32545.
14. Contag CH, Spilman SD, Contag PR, et al. Visualizing gene
expression in living mammals using a bioluminescent reporter.
Photochem Photobiol 1997;66(4):52331.
15. Edinger M, Sweeney TJ, Tucker AA, Olomu AB, Negrin RS,
Contag CH. Noninvasive assessment of tumor cell proliferation in
animal models. Neoplasia 1999;1(4):30310.
16. Paroo Z, Bollinger RA, Braasch DA, et al. Validating bioluminescence imaging as a high-throughput, quantitative modality for
assessing tumor burden. Mol Imaging 2004;3(2):11724.
17. Burgos JS, Rosol M, Moats RA, et al. Time course of bioluminescent signal in orthotopic and heterotopic brain tumors in nude mice.
Biotechniques 2003;34(6):11848.
18. Wu JC, Sundaresan G, Iyer M, Gambhir SS. Noninvasive optical
imaging of firefly luciferase reporter gene expression in skeletal
muscles of living mice. Molec Ther 2001;4(4):297306.
19. Wang W, El-Deiry WS. Bioluminescent molecular imaging of
endogenous and exogenous p53-mediated transcription in vitro
and in vivo using an HCT116 human colon carcinoma xenograft
model. Cancer Biol Ther 2003;2(2):196202.
20. Avram MJ, Krejcie TC, Niemann CU, Enders-Klein C, Shanks CA,
Henthorn TK. Isoflurane alters the recirculatory pharmacokinetics
of physiologic markers. Anesthesiology 2000;92(6):175768.
21. Baba S, Cho SY, Ye Z, Cheng L, Engles JM, Wahl RL. How
reproducible is bioluminescent imaging of tumor cell growth?
Single time point versus the dynamic measurement approach. Mol
Imaging 2007;6(5):31522.
22. Edinger M, Cao YA, Hornig YS, et al. Advancing animal models
of neoplasia through in vivo bioluminescence imaging. Eur J
Cancer 2002;38(16):212836.
23. Sarraf-Yazdi S, Mi J, Dewhirst MW, Clary BM. Use of in vivo
bioluminescence imaging to predict hepatic tumor burden in mice.
J Surg Res 2004;120(2):24955.
24. Scatena CD, Hepner MA, Oei YA, et al. Imaging of bioluminescent LNCaP-luc-M6 tumors: a new animal model for the study of
metastatic human prostate cancer. Prostate 2004;59(3):292303.