Impact of dynamic 18F-FDG PET on the early prediction of therapy outcome in patients with high-risk soft-tissue sarcomas after neoadjuvant chemotherapy: a feasibility study.

Abstract

Dynamic PET (dPET) studies with (18)F-FDG were performed in patients with soft-tissue sarcomas who received neoadjuvant chemotherapy early in the course of therapy. The goal of the study was to evaluate the impact of early dPET studies and assess their value with regard to the therapy outcome using histopathologic data.
The evaluation included 31 patients with nonmetastatic soft-tissue sarcomas, who were treated with neoadjuvant chemotherapy consisting of etoposide, ifosfamide, and doxorubicin. Patients were examined before the onset of therapy and after the completion of the second cycle. Histopathologic response served for reference and was available for 25 of 31 patients. Response was defined as less than 10% viable tumor tissue in the resected tumor tissue. The following parameters were retrieved from dPET studies: standardized uptake value (SUV); fractal dimension; 2-compartment model with computation of K(1), k(2), k(3), and k(4) (unit, 1/min); fractional blood volume; and influx according to Patlak.
The mean SUV was 4.6 before therapy and 2.8 after 2 cycles. The mean influx was 0.059 before therapy and 0.043 after 2 cycles. The mean SUV was 3.9 in the responders and 5.5 in the nonresponders before therapy. After therapy, responders revealed a mean SUV of 2.5, whereas nonresponders had a mean SUV of 3.5. We used linear discriminant analysis to categorize the patients into 2 groups: response (n = 12) and nonresponse (n = 13). The correct classification rate of the responders (positive predictive value) was generally higher (>67%) than that for the nonresponders. Finally, the combined use of the 2 predictor variables, namely SUV and influx, of each study led to the highest accuracy of 83%. This combination was particularly useful for the prediction of responders (positive predictive value, 92%). The use of the percentage change in maximum SUV led to an accuracy of 58%.
On the basis of these results, only a multiparameter analysis based on kinetic (18)F-FDG data of a baseline study and after 2 cycles is helpful for the early prediction of chemosensitivity in patients with soft-tissue sarcomas receiving neoadjuvant chemotherapy.

Full text

Page 1

Impact of Dynamic18F-FDG PET on the
Early Prediction of Therapy Outcome in
Patients with High-Risk Soft-Tissue Sarcomas
After Neoadjuvant Chemotherapy: A
Feasibility Study
Antonia Dimitrakopoulou-Strauss1, Ludwig G. Strauss1, Gerlinde Egerer2, Julie Vasamiliette1,
Gunhild Mechtersheimer3, Thomas Schmitt2, Burkhard Lehner4, Uwe Haberkorn1,5,
Philipp Stroebel6, and Bernd Kasper2
1Medical PET Group–Biological Imaging, Clinical Cooperation Unit Nuclear Medicine, German Cancer Research Center,
Heidelberg, Germany;2Department of Internal Medicine V, University of Heidelberg, Heidelberg, Germany;3Institute of Pathology,
University of Heidelberg, Heidelberg, Germany;4Orthopedic University Hospital of Heidelberg, Heidelberg, Germany;5Division of
Nuclear Medicine, University of Heidelberg, Heidelberg, Germany; and6Institute of Pathology, University Medical Center
Mannheim, University of Heidelberg, Heidelberg, Germany
Dynamic PET (dPET) studies with18F-FDG were performed in
patients with soft-tissue sarcomas who received neoadjuvant
chemotherapy early in the course of therapy. The goal of the
study was to evaluate the impact of early dPET studies and as-
sess their value with regard to the therapy outcome using histo-
pathologic data. Methods: The evaluation included 31 patients
with nonmetastatic soft-tissue sarcomas, who were treated
with neoadjuvant chemotherapy consisting of etoposide, ifosfa-
mide,anddoxorubicin. Patients were examinedbefore the onset
of therapy and after the completion of the second cycle. Histo-
pathologic response served for reference and was available for
25of 31 patients. Response was defined asless than10%viable
tumor tissue in the resected tumor tissue. The following param-
eters were retrieved from dPET studies: standardized uptake
value (SUV); fractal dimension; 2-compartment model with com-
putationofK1,k2,k3,andk4(unit,1/min);fractionalbloodvolume;
and influx according to Patlak. Results: The mean SUV was 4.6
before therapy and 2.8 after 2 cycles. The mean influx was 0.059
before therapy and 0.043 after 2 cycles. The mean SUV was 3.9
in the responders and 5.5 in the nonresponders before therapy.
After therapy, responders revealed a mean SUV of 2.5, whereas
nonresponders had a mean SUV of 3.5. We used linear discrim-
inant analysis to categorize the patients into 2 groups: response
(n512)andnonresponse(n513).Thecorrectclassificationrate
ofthe responders(positive predictive value)was generally higher
(.67%) than that for the nonresponders. Finally, the combined
use of the 2 predictor variables, namely SUV and influx, of
eachstudy led to the highest accuracy of 83%.Thiscombination
was particularly useful for the prediction of responders (positive
predictive value, 92%). The use of the percentage change in
maximum SUV led to an accuracy of 58%. Conclusion: On the
basis of these results, only a multiparameter analysis based on
kinetic18F-FDG data of a baseline study and after 2 cycles is
helpful for the early prediction of chemosensitivity in patients
with soft-tissue sarcomas receiving neoadjuvant chemotherapy.
KeyWords:18F-FDG;sarcomas;kineticmodeling;therapymon-
itoring; prognosis
J Nucl Med 2010; 51:551–558
DOI: 10.2967/jnumed.109.070862
Soft-tissue sarcomas (STS) are a heterogeneous group
of connective-tissue tumors that commonly arise in the
extremities and constitute less than 1% of all adult
malignancies (1). Sarcoma-related mortality from extremity
lesions occurs most commonly secondary to hematogenous
metastasis (2). Curative treatment of STS requires complete
surgical resection of the primary tumor, which is not always
possible. Combined treatment protocols including neo-
adjuvant chemotherapy (NAC) are in use in many centers as
a part of a treatment plan, particularly in patients with
primary STS in the extremities. The theoretic consider-
ations for using NAC are earlier treatment of microscopic
metastatic disease and facilitation of tumor removal. How-
ever, the value of neoadjuvant treatment remains unproven.
Furthermore, it is known that only a subcollective of
patients benefits from NAC; therefore, there is an urgent
need for sensitive methods, which allow an early iden-
tification of NAC responders.
Received Sep. 18, 2009; revision accepted Dec. 10, 2009.
For correspondence or reprints contact: Antonia Dimitrakopoulou-
Strauss, Medical PET Group–Biological Imaging (E0601), Clinical
Cooperation Unit Nuclear Medicine, German Cancer Research Center,
Im Neuenheimer Feld 280, D-69120 Heidelberg, Germany.
E-mail: ads@ads-lgs.de or a.dimitrakopoulou-strauss@dkfz.de
COPYRIGHT ª 2010 by the Society of Nuclear Medicine, Inc.
18F-FDG MONITORING IN SARCOMAS • Dimitrakopoulou-Strauss et al.551

Page 2

PET with18F-FDG is being used increasingly, primarily
in patients with solid tumors for staging and therapy
monitoring. The idea is that the reduction in
uptake after chemotherapy indicates a response and a better
patient outcome and that, beyond this, the change in18F-
FDG uptake is a more sensitive criterion for the change in
tumor volume. However, it is known that the use of18F-FDG
PET for the assessment of therapeutic response is a complex
topic that is still under evaluation because of the lack of
widely accepted standardized protocols concerning the data
acquisition and evaluation. The results of monitoring studies
may vary, depending on the methodology used and the
selection of the time point for the follow-up study.
The purpose of this study was to examine whether
a baseline or an early18F-FDG follow-up study after the
completion of the second cycle of a combined chemother-
apy with etoposide, ifosfamide, and doxorubicin (Adria-
mycin; Pfizer Pharma) (EIA) is helpful for the prediction of
the therapy outcome of NAC. Furthermore, another aim
was to establish a methodology that can be used for the
18F-FDG
selection of those patients who may benefit from EIA
therapy early in the treatment course, given the fact that
only a portion of patients responds to NAC. The compar-
ison of the PET18F-FDG data to clinical risk parameters
was not topic of this article.
MATERIALS AND METHODS
This prospective study includes 31 patients with primary non-
metastatic histologically proven STS. All patients received com-
bined NAC consisting of etoposide (125 mg/m2on days 1 and 4),
ifosfamide (1,500 mg/m2on days 1 through 4), and doxorubicin
(50 mg/m2on day 1) (EIA regimen, 4–6 cycles). The included
patients were not diabetic and had not had previous chemotherapy.
All patients were examined before the onset of chemotherapy and
after the completion of the second cycle. Patients gave written
informed consent to participate in the study and to have their
medical records released. The study was approved by the Ethical
Committee of the University of Heidelberg.
The histopathologic data were correlated to the PET data.
Relevant patient data are demonstrated in Table 1. Histopathologic
response data were available in 25 of 31 patients. Three patients
TABLE 1. Patients’ Characteristics
Patient
no.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Histology and grade
Synovial sarcoma, II
Leiomyosarcoma, III
Synovial sarcoma, III
Liposarcoma, III
Fibrosarcoma, III
Fibrosarcoma, II
Fibrosarcoma, II
Synovial sarcoma, II
Liposarcoma, II
Liposarcoma, III
Epitheloid sarcoma, III
Liposarcoma, III
MFH, III
Liposarcoma, II
Rhabdomyosarcoma, II
MFH, II
Haemangiopericytoma, II
Liposarcoma, III
Liposarcoma, III
Liposarcoma, II
MFH, III
Synovial sarcoma, III
Synovial sarcoma, III
Leiomyosarcoma, III
Synovial sarcoma, III
Liposarcoma, II
Liposarcoma, II
Synovial sarcoma, III
Liposarcoma, III
MFH, III
Liposarcoma, II
Lesion location
in PET
Upper leg
Upper leg
Mediastinum
Retroperitoneum
Upper leg
Upper leg
Upper leg
Upper arm
Upper leg
Thyroid gland
Arm/shoulder
Thoracic wall
Thoracic wall
Upper leg
Sinus maxillaris
Lower leg
Pelvic area
Upper leg
Retroperitoneum
Upper leg
Upper leg
Pelvic area
Upper leg
Thoracic wall
Upper leg
Upper leg
Upper leg
Lower leg
Upper leg
Upper leg
Upper leg
Cycles
of EIA
4
4
6
6
4
2
4
4
5
4
4
4
4
4
4
5
8
4
4
4
4
4
4
4
4
4
4
4
2
4
4
Resection
before
EIA
R1
R1
R2
R2
R1
R2
No
R1
No
R2
No
R0
No
No
No
No
No
No
No
No
No
No
R2
R0–1
No
No
No
No
No
No
No
Response
Evaluation
Criteria in
Solid Tumors
CR
PR
SD
SD
CR
SD
SD
CR
PR
PR
PR
SD
SD
SD
PR
PR
SD
SD
SD
SD
SD
SD
PR
PD
SD
SD
SD
SD
PD
SD
SD
Histopathologic
response (,10%
viable cells)
Response
Response
No surgery
Response
Response
Nonresponse
NA
Response
Response
Response
Response
No surgery
Nonresponse
Nonresponse
Response
Nonresponse
No surgery
Nonresponse
Nonresponse
Nonresponse
Nonresponse
Nonresponse
Response
Nonresponse
NA
Nonresponse
Response
Nonresponse
Response
NA
Nonresponse
Treatment
after EIA
R0 1 IORT
R0 1 IORT
RTx
R1 1 IORT
R0 1 IORT
R0
R0 1 IORT
R0 1 IORT
R0
R1
R0 1 RTx
No surgery
R0 1 IORT
R1 1 RTx
R1
R1
No surgery
R0 1 IORT
R1 1 IORT
R0 1 IORT
Rx 1 IORT
R1 1 IORT
R0 1 IORT
R0 1 IORT
R0 1 IORT
R0 1 IORT
R0 1 IORT
R0 1 IORT
R0 1 IORT
R0 1 IORT
R0 1 IORT
R1 5 R1 resection; CR 5 complete responder; R0 5 R0 resection; IORT 5 intraoperative radiation treatment; PR 5 partial
responder; R2 5 R2 resection; SD 5 stable disease; RTx 5 radiation treatment; NA 5 not applicable; MFH 5 malignant fibrous
histocytoma; Rx 5 Rx resection; PD 5 progressive disease.
552THE JOURNAL OF NUCLEAR MEDICINE • Vol. 51 • No. 4 • April 2010

Page 3

refused surgery. Response was defined as less than 10% viable
tumor tissue in the resected tumor tissue. The histologic tumor
response was assessed using the Salzer-Kuntschik classification
(3). This classification system was initially developed to grade
therapy-induced tumor regression of osteosarcomas but has sub-
sequently also been used for soft-tissue tumors. The classification
is based on a semiquantitative scoring of viable and necrotic tumor
in a given sample. Regression grade I is defined as no viable
tumor, grade II as single viable tumor cells only, grade III as less
than 10% viable tumor, grade IV as 10%250% viable tumor,
grade V as more than 50% viable tumor, and grade VI as 100%
viable tumor. The grade classification of tumor response was
determined by a pathologist. Histologic response was defined as
less than 10% viable tumor tissue in the resected tumor tissue
(corresponding to grades I–II).
Dynamic PET (dPET) studies were performed over the primary
tumor for 60 min after the intravenous application of 300–370 MBq
of18F-FDG using a 28-frame protocol (10 frames of 30 s, 5 frames
of 60 s, 5 frames of 120 s, and 8 frames of 300 s). Additional static
images were acquired in all patients by moving the table in cranial
and caudal positions in relation to the initial position. The gap
used for each repositioning was 13.5 cm. A dedicated PET system
(ECAT EXACT HR1; Siemens Co.) with an axial field of view of
15.3 cm, operated in septa-extended (2-dimensional) mode, was
used for patient studies. The system allowed the simultaneous
acquisition of 63 transversal slices with a theoretic slice thickness
of 2.4 mm. For attenuation correction of the acquired dynamic
emission tomographic images, a transmission scan for a total of 10
min was obtained before radionuclide application. A transmission
scan of 3 min, followed by an emission scan of 7 min, was
acquired for each additional static image. All PET images were
attenuation-corrected, and an image matrix of 256 · 256 pixels
was used for iterative image reconstruction. The reconstructed
images were converted to standardized uptake value (SUV)
images (4). The SUV 55–60 min after injection was used for
the analysis of the18F-FDG uptake.
dPET data were evaluated using the software package PMod
(PMOD Technologies Ltd.) (5,6). Two experienced nuclear
medicine physicians visually analyzed the hypermetabolic areas
on transaxial, coronal, and sagittal images.
All patients had a diagnostic MRI before the first PET study.
Generally, the quantitative evaluation was based on irregular
volumes of interest (VOIs) placed over the tumor and an arterial
vessel within the field of view and in reference tissue (mostly the
muscle in the contralateral side). For the better delineation of the
tumor tissue and the improved detection of hypermetabolic
lesions, parametric images were calculated on the basis of the
dPET data by fitting a linear regression function to the time–
activity data for each voxel. Details of this method have been
described elsewhere (7). Parametric images of the distribution
volume have been used for better anatomic delineation of the
vessels. Time–activity curves were created using VOIs. A VOI
consisted of several regions of interest (ROIs) over the target area.
Irregular ROIs were drawn manually. To compensate for possible
patient motion during the acquisition time, the original ROIs were
visually repositioned but not redrawn. A detailed quantitative
evaluation of tracer kinetics requires the use of compartment
modeling. A 2-tissue-compartment model is the standard meth-
odology for the quantification of dynamic18F-FDG studies (8,9).
One problem in patient studies is the accurate measurement of
the input function, which theoretically requires arterial blood
sampling. However, the input function can be retrieved from the
image data with good accuracy (10). For the input function, the
mean value of the VOI data obtained from a large arterial vessel
was used. A vessel VOI consisted of at least 7 ROIs in sequential
PET images. The recovery coefficient is 0.85 for a diameter of
8 mm and for the system described above. Partial-volume correc-
tion was performed in selected cases for small vessels or lesions
(diameter, ,8 mm) on the basis of phantom measurements of the
recovery function using the PMod software. Noise in the input
curve affects the parameter estimates. Therefore, we used a pre-
processing tool, available in the PMod software, that allowed
a fit of the input curve, namely by a sum of up to 3 decaying
exponentials to reduce noise. The constants K1–k4were calculated
using a 2-compartment model implemented in the PMod software,
taking into account the vascular fraction (VB) in a VOI as an
additional variable. Details about the applied compartment model
are described by Burger and Buck (5).
One major advantage of the PMod software is the graphical
interface that allowed the interactive configuration of the kinetic
model by the user and the application of some preprocessing steps,
for example, setting up initial values and limits for the fit
parameters. Each plot was evaluated visually to check the quality
of fit. Each model curve was compared with the corresponding
time–activity curve, and the total x2difference was used as the
cost function, for which the criterion was to minimize the summed
squares (x2) of the differences between the measured and the
model curve (x2was usually ,2). This means that the squared
residuals (measured value minus estimated value) are multiplied
by weights. In theory, the weight should be related to the SE of
a measurement. The distribution at each individual point is taken
to be gaussian, with an SD to be specified. The residual covariance
was dependent on the kinetic parameter and typically less than
10% for K1. The model parameters were accepted when VB and
K1–k4were less than one and the VB values exceeded zero. The
unit for the rate constants K1–k4is 1/min, and VB reflects the
fraction of blood within the evaluated volume. After compartment
analysis, we calculated the global influx of18F-FDG from the
compartment data using the formula influx 5 (K1· k3)/(k21 k3).
In addition to the compartment analysis, a noncompartment
model based on the fractal dimension (FD) was used. The FD is
a parameter for the heterogeneity and was calculated for the time–
activity data in each individual voxel of a VOI. The values of the
FD vary from 0 to 2, showing the deterministic or chaotic
distribution of the tracer activity. We used a subdivision of 7 ·
7 and a maximal SUV of 20 for the calculation of FD (11).
Parametric images were calculated using dPET data by fitting
a linear regression function to the time–activity data and for each
pixel. Images of the slope and the intercept of the time–activity data
have been calculated using PMod software. Parametric images of
the slope reflect the trapping of18F-FDG and have been used for the
delineation of the malignant lesions and the VOIs placement
because of high contrast to the surrounding tissue. Parametric
images of the intercept reflect the distribution volume of18F-FDG
and have been used for the better anatomic localization of the
lesions because of the delineation of the vessels.
Data were statistically evaluated using the STATA/SE 10.1
(StataCorp) software on a Quad-Core Intel Xeon (2 · 3.0 GHz, 16
GB RAM) running with Mac OS X 10.5.8 (Apple Computer
International). The statistical evaluation was performed using the
descriptive statistics, box plots, and Wilcoxon matched-pairs
signed rank test. The results were considered significant for
18F-FDG MONITORING IN SARCOMAS • Dimitrakopoulou-Strauss et al.553

Page 4

P less than 0.1. Linear discriminant analysis with equal prior
probabilities has been applied to the data.
RESULTS
The patient characteristics and individual survival data
are presented in Table 1. Twenty-one patients had a primary
tumor in the extremities (n 5 17 in the upper legs, n 5 2 in
the lower legs, and n 5 2 in the upper arms/shoulder), 1
patient in the mediastinum, 2 patients in the retroperitoneal
area, 1 patient in the thyroid gland, 3 patients in the thoracic
wall and axilla, 1 patient in the sinus maxillaris, and 2
patients in the pelvic area. Eighteen of the patients had
a grade III tumor, and the remaining 13 patients had a grade
II tumor. The diagnostic examinations performed before
PET were MRI or CT. The focal18F-FDG findings were
correlated to the external CT or MRI findings using a site-
by-site analysis.
In 3 of 31 patients, surgery could not be performed. Of
the remaining 28 patients, 23 patients received a surgical
resection, followed by radiation therapy, in most cases
intensity-modulated radiation therapy. Five patients refused
adjuvant radiation therapy.
All tumors demonstrated an enhanced18F-FDG uptake in
the baseline study. The evaluation of the follow-up18F-
FDG study was in some cases more difficult than that of the
baseline study. To delineate the tumor and to use VOIs com-
parable to those used in the baseline study, we imported the
VOIs of the baseline study and adjusted the original VOIs
to the follow-up study. Parametric images based on the
regression function have been used for the placement of
VOIs and for both studies, to better delineate the tumors
and evaluate the tumor areas with the highest metabolic
activity. For this purpose, the slope images have been used.
Figure 1 demonstrates an example of a patient with
a grade II liposarcoma of the left upper leg, who did not
respond to NAC (4 cycles EIA). The kinetic data revealed
a decrease in mean SUV from 12.76 to 3.33 (73%), a
decrease in k3from 0.104 to 0.037, and almost constant
values for K1 (baseline, 0.9; follow-up, 0.8) and VB
(baseline, 0.132; follow-up, 0.085). The restaging data after
4 cycles EIA indicated stable disease; the histopathologic
data, however, revealed no response to EIA (viable tumor
cells . 90%).
Figure 2 demonstrates an example of a patient with a
grade III epitheloid sarcoma in the right upper arm/shoulder,
who responded to NAC (4 cycles EIA). The kinetic data
revealed a decrease in mean SUV from 9.04 to 1.63 (82%),
a decrease in k3(baseline, 0.161; follow-up, 0.099), and
a decrease in K1(baseline, 0.442; follow-up, 0.084) and VB
(baseline, 0.113; follow-up, 0.044). The restaging data after
4 cycles of EIA indicated a partial remission, and the
histopathologic data revealed a response to EIA (,5%
viable tumor tissue).
Comparison Between First and Second Studies
The mean SUV was 4.6 before therapy and 2.8 after
2 cycles in all evaluated patients (n 5 31) (Supplemental
Table 1a; supplemental materials are available online only
at http://jnm.snmjournals.org). The mean influx was 0.059
before therapy and 0.043 after 2 cycles (n 5 31). The mean
SUV was 3.9 in the group of responders (n 5 12) and 5.5 in
the group of nonresponders (n 5 13) before therapy.
Responders revealed a mean SUV of 2.5 after therapy, as
compared with 3.5 SUV for nonresponders.
Most kinetic parameters demonstrated only small changes,
typically declining after 2 chemotherapeutic cycles. The
changes in the kinetic parameters were analyzed by the
Wilcoxon matched-pairs signed rank test. The test revealed
a significant change for K1(P 5 0.0735), k4(P 5 0.0734),
and FD (P 5 0.0941) when the absolute values of the first
and the second studies were compared (Supplemental Table
1a). The comparison of the kinetic parameters between the
first and the second PET studies with respect to histopath-
ologic response (n 5 25) revealed statistically significant
differences only for the study after therapy for K1(P 5
0.0735), k4(P 5 0.0734), and FD (P 5 0.0941) (Supple-
mental Table 1b). Figures 3, 3A23C demonstrate box-and-
whisker plots of the median values for all kinetic parameters
before and after therapy in responders and nonresponders.
FIGURE 1.
therapy in patient with grade II liposarcoma in medioventral
part of upper leg, and corresponding18F-FDG image of
same patient after 2 cycles of EIA. Histopathologic data
revealed no response to EIA (.90% viable tumor cells).
Transversal
18F-FDG PET image before EIA
FIGURE 2.
therapy in patient with grade III epitheloid sarcoma in right
upper arm/shoulder, and corresponding18F-FDG image of
same patient after 2 cycles of EIA. Histopathologic data
revealed a response to EIA (,5% viable tumor tissue).
Transversal
18F-FDG PET image before EIA
554THE JOURNAL OF NUCLEAR MEDICINE • Vol. 51 • No. 4 • April 2010

Page 5

Group-Based Classification According to
Histopathologic Data by Use of Discriminant Analysis
The histopathologic response was classified as less than
10% viable tumor tissue. We categorized the patients (n 5
25) according to this criterion and defined 2 response
groups. Table 2 presents the results of the linear discrim-
inant analysis with equal prior probabilities for each study
and their combinations. The data demonstrate that the clas-
sification was generally higher (positive predictive value
[PPV] . 67%) for the responders than for the non-
responders. The use of mean SUV and influx value was,
in particular, high for the group of responders (PPV, 92%).
Overall, the nonresponders could be best classified using
the mean SUVand influx or the FD and k4of the follow-up
study (negative predictive value, 83%). Finally, the com-
bined use of the 2 predictor variables, namely mean SUV
and influx, of each study led to the highest accuracy, 83%,
for both groups. This combination was particularly useful
for the prediction of responders (PPV, 92%). The use of the
percentage change in maximum SUV (SUVmax) led to
a sensitivity of 57%, a specificity of 60%, and an accuracy
of 58%.
DISCUSSION
18F-FDG imaging greatly affects the diagnostics and
management of oncologic patients, in particular those who
have a solid tumor. The impact of the predictive value of
18F-FDG PET early in the course of chemotherapy is still
open and under evaluation. Only a few reports exist on
therapy monitoring of NAC in sarcomas using18F-FDG
PET. The main difference between these studies and ours is
the early time chosen for the follow-up in our design,
consistent scheme for NAC, and data evaluation. The pri-
mary clinical demand is to assess the cytostatic effect early
in the course of NAC to exclude nonresponders or modify
the therapeutic protocol accordingly. Benz et al. reported
on a study with 20 patients with locally advanced high-
grade STSs who had been studied with18F-FDG PET/CT
before and after the completion of the preoperative therapy,
which included different chemotherapeutic protocols; 70%
of the patients underwent additional external-beam radia-
tion (12). The authors reported significant differences in
SUV changes (mean and maximum) in histopathologic
responders (70%278%), as compared with the nonre-
sponders (27%240%). Histopathologic response was de-
fined as less than 5% viable tumor tissue. In contrast, the
changes in tumor volume as measured by CT did not allow
the prediction of response. One limitation of this study is
that only 6 of 20 patients were responders, which may raise
statistical problems regarding the accuracy. Evilevitch et al.
reported on a similar study including 42 patients with
resectable high-grade sarcomas who received combined
NAC and underwent18F-FDG PET/CT before and after the
completion of NAC (13). Also in this study, NAC was not
consistent, and 57% of the patients received additional
external-beam radiation. The authors reported on a greater
FIGURE 3.
mean SUV 55–60 min after injection before and after 2
cycles of NAC in responders and nonresponders. (B) Box-
and-whisker plots of median tumor values for VB, K1, and k3
before and after 2 cycles of NAC in responders and
nonresponders. VB does not have units; K1and k3values
are expressed in 1/min. (C) Box-and-whisker plots of median
tumor FD before and after 2 cycles of NAC in responders
and nonresponders.
(A) Box-and-whisker plots of median tumor
18F-FDG MONITORING IN SARCOMAS • Dimitrakopoulou-Strauss et al.555

Page 6

reduction in18F-FDG uptake in histopathologic responders,
whereas no significant changes were found for the tumor
size. The best cutoff level for the assessment of histopath-
ologic response was 60% decrease in tumor
uptake, which led to a sensitivity of 100% and a specificity
of 71%. The results demonstrate that a simple cutoff level
for the change in SUV is related to low specificity. In
contrast, Response Evaluation Criteria in Solid Tumors
provided a sensitivity of 25% and a specificity of 100%. In
our study, sensitivity was 78.6% and specificity 90% when
using the absolute values of mean SUV and influx of both
studies, compared with 64% and 70% when using only the
mean SUVs of both studies. In particular, PPV (92%) and
accuracy (83%) were highest when using the combination
of mean SUV and influx of both studies. The data support
the use of dPET, which is a prerequisite for the calculation
of kinetic data for the prediction of therapy outcome.
Because early measurements within NAC in sarcoma
patients—as compared with measurements within a longer
time interval and after several chemotherapeutic cycles in
sarcoma patients—deal with smaller changes in the18F-
FDG metabolism, the use of dPET and calculation of
kinetic data seem to be helpful.
Some authors reported good results in other tumor types
by performing a baseline and 1 follow-up study early in the
course of NAC, within the first cycle. A reduction in18F-
FDG uptake early in the course of NAC (14 d after onset)
was a good prognostic criterion for the histopathologic
response classification of tumors in the esophagogastric
junction and was characterized by significantly longer
time to progression and longer overall survival (14). The
reported sensitivity was 89% and specificity was 75% with
respect to the histopathologic response using a cutoff value
of 35% reduction in18F-FDG metabolism, as compared
with 57% sensitivity and 60% specificity for the percentage
change in SUVmax, in our study. This may be related to the
different histology and chemotherapeutic protocol used.
The literature data concerning the change in
uptake after chemotherapy and long-term therapy outcome
are divergent. Schuetze et al. demonstrated in patients with
18F-FDG
18F-FDG
localized extremity STSs treated with NAC that less than
40% decrease in SUVmax was a bad prognostic criterion,
because these patients were at high risk of systemic dis-
ease recurrence estimated to be 90% at 4 y from the initial
diagnosis (15). The main problem of the use of percentage
change of
18F-FDG metabolism is that sensitivity and
specificity depend on the cutoff level used for discrimi-
nation, and dependence on cutoff level is furthermore
dependent on the tumor histology and therapy used. The
greater the percentage of18F-FDG reduction, the better
the therapeutic result, as shown in different studies
(16,17). Cerfolio et al. recommended the use of a decrease
of more than 80% in SUVmax for the prediction of
complete pathologic response, with a sensitivity of 90%,
specificity of 100%, and accuracy of 96% in patients with
non–small cell lung tumors who received NAC (18). In
our study, the change in SUVmax led to a sensitivity of
57% and a specificity of 60% and was worse than the
combination of the absolute mean SUV and influx values
of both studies.
Most of the studies about monitoring the chemothera-
peutic effect using PET in oncologic patients are based on
simple quantification methods for data analysis. Most
authors used only a visual or a semiquantitative method,
for example, SUVor tumor–to–normal-tissue ratio. The use
of SUV for the quantification of18F-FDG studies is a robust
parameter that reflects tumor viability and has a prognostic
value. Nonresponders demonstrated higher mean SUVs
in the baseline study (5.5 vs. 3.9) as shown in the present
study (Supplemental Table 1a) and as reported in the liter-
ature (8,12,15). Therefore, baseline18F-FDG uptake may
have a prognostic value. Schuetze et al. demonstrated in
patients with localized extremity STSs treated with NAC
that a pretreatment tumor SUVmax greater than 6 was a
bad prognostic criterion because these patients were at high
risk of systemic disease recurrence estimated to be 90% at
4 y from the initial diagnosis (15). We reported that higher
baseline18F-FDG metabolism as expressed in mean SUVor
k3or FD in patients with non–small cell lung cancer was
related to a shorter progression-free survival after chemo-
TABLE 2. Results of Linear Discriminant Analysis with Equal Prior Probabilities Based on18F-FDG Parameters of First
PET Study (1) or Second PET Study (2) or Combination of Both Studies
Parameter Sensitivity
9/15 (60.00%)
9/11 (81.81%)
10/16 (62.5%)
8/9 (88.90%)
9/11 (81.81%)
9/14 (64.30%)
11/14 (78.60%)
8/14 (57.14%)
Specificity
7/10 (70.00%)
3/14 (21.43%)
6/8 (75.00%)
6/10 (60.00%)
3/13 (23.00%)
7/10 (70.00%)
9/10 (90.00%)
6/10 (60.00%)
PPVNPVAccuracy
16/25 (64.00%)
19/25 (76.00%)
16/24 (66.70%)
18/24 (75.00%)
19/24 (79.20%)
16/24 (66.70%)
20/24 (83.33%)
14/24 (58.33%)
1: SUV
1: SUV, VB, k1, k3, FD
2: SUV
2: SUV, influx
2: FD, k4
1 1 2: SUV
1 1 2: SUV, influx
% change SUVmax
9/12 (75.00%)
9/12 (75.00%)
10/12 (83.33%)
8/12 (67.00%)
9/12 (75.00%)
9/12 (75.00%)
11/12 (91.67%)
8/12 (66.67%)
7/13 (54.00%)
10/13 (77.00%)
6/12 (50.00%)
10/12 (83.30%)
10/12 (83.30%)
7/12 (58.33%)
9/12 (75.00%)
6/12 (50.00%)
Groups were defined according to histologic classification of 10% variable tumor tissue.
PPV 5 positive predictive value; NPV 5 negative predictive value.
556THE JOURNAL OF NUCLEAR MEDICINE • Vol. 51 • No. 4 • April 2010

Page 7

therapy (7). In patients with colorectal liver metastases,
baseline mean SUVs greater than 6 were associated with an
overall survival of less than 200 d after cytostatic treatment
(19). The question is, however, how reliable is baseline
mean SUV or its change for a response classification?
Larson et al. proposed the use of another semiquantitative
index of total tumor glycolysis (TLG) for better describing
response to treatment (20). TLG is calculated by multiply-
ing the CT volume and the SUV. However, because of the
relatively small changes of tumor volume related to the
presence of necrosis, edema, or hemorrhage—in particular
early after chemotherapy—the use of TLG for early moni-
toring is doubtful as demonstrated in the study of Benz
et al. (12). The authors demonstrated that the use of
TLG diminished the histopathologic response information.
Therefore, the use of TLG cannot be recommended on the
basis of the literature data available.
Full dynamic quantitative studies provide the possibility
of extending quantification and getting data about the18F-
FDG kinetics. The hypothesis is that different time–activity
curves of18F-FDG can achieve the same SUVas measured
by a single static measurement 55–60 min after injection.
However, by applying compartmental and noncompartmen-
tal analysis to the dynamic data, the shape of the time–
activity curve is characterized, and therefore additional
valuable information concerning the distribution volume
VB (which is related to perfusion), influx and efflux of the
tracer, and phosphorylation and dephosphorylation rate can
be obtained. According to the results of the present study,
the influx data were helpful, particularly in combination
with SUV for the correct response classification. The cal-
culation of influx comprises K1, k2, and k3, which are the
major compartment parameters. One problem that should
be discussed is whether the use of just 1 parameter of the
18F-FDG kinetics, such as the transport rates, is enough for
an accurate prediction of response. Although nonre-
sponders demonstrated higher mean SUV, FD, VB, and
K1–k3values before therapy (Supplemental Table 1b), the
difference was not statistically significant. In the follow-up
study, we found significant differences at a level of P less
than 0.1 for K1, k4, and FD. Nonresponders were related to
higher influx values, lower phosphorylation rates, and
higher FD values after NAC. However, the presented data
demonstrate that the best results for a group-based analysis,
into responders and nonresponders, were obtained on the
basis of a multiparameter analysis including the absolute
values of mean SUV and influx of the baseline study and
the follow-up study. The data demonstrate that EIA therapy
may have a combined effect on both the angiogenesis and
the proliferation because of the more pronounced effect
on SUV and influx, which is a parameter that takes into
account K1, k2, and k3. In particular, the prediction of
response was better (92%) based on the combination of the
kinetic data, in comparison to the use of mean SUV alone
(75%). The use of the percentage change of SUVmax
cannot be recommended on the basis of the results of this
study. Another major limitation of the use of SUVmax is
that it is highly dependent on the statistical quality of the
images and the size of the maximal pixel and is, therefore,
less robust than the use of the average SUV within a VOI
(21). The data give evidence for an assessment of the
inhibitory effects of this type of chemotherapy using full
kinetic analysis of the18F-FDG data of the baseline study
and the study after the completion of 2 cycles of NAC.
CONCLUSION
Prediction of preoperative response is a topic that raises
several questions concerning the handling of the data. The
data of this study demonstrate that only a multiparameter
analysis including the combination of the mean absolute
values of mean SUV and influx of a baseline study and
a follow-up study after completion of 2 cycles of NAC is a
robust combination for a group-based analysis (of catego-
rizing patients into response or nonresponse). The quanti-
tative assessment of the18F-FDG kinetics in tumors should
be used to measure the inhibitory effect of the che-
motherapy on the tumor growth. In particular, the presented
methodology may be used in patients with high-risk STSs,
because proper selection of those patients who may benefit
from a neoadjuvant chemotherapeutic treatment early in the
course of treatment is necessary.
REFERENCES
1. National Cancer Institute. Soft Tissue Sarcomas: Questions and Answers.
Available at: http://www.cancer.gov/cancertopics/factsheet/sites-types/soft-tissue-
sarcoma. Accessed January 25, 2010.
2. Grobmyer SR, Maki RG, Demetri GD, et al. Neo-adjuvant chemotherapy for
primary high-grade extremity soft tissue sarcoma. Ann Oncol. 2004;15:1667–
1672.
3. Salzer-Kuntschik M, Brand G, Delling G. Determination of the degree of
morphological regression following chemotherapy in malignant bone tumors.
Pathologe. 1983;4:135–141.
4. Strauss LG, Conti PS. The applications of PET in clinical oncology. J Nucl Med.
1991;32:623–648.
5. Burger C, Buck A. Requirements and implementations of a flexible kinetic
modeling tool. J Nucl Med. 1997;38:1818–1823.
6. Mikolajczyk K, Szabatin M, Rudnicki P, Grodzki M, Burger C. A Java
environment for medical image data analysis: initial application for brain PET
quantitation. Med Inform (Lond). 1998;23:207–214.
7. Dimitrakopoulou-Strauss A, Hoffmann M, Bergner R, et al. Prediction of short-
term survival in patients with advanced nonsmall cell lung cancer following
chemotherapy based on 2-deoxy-2-(F-18)-fluoro-D-glucose positron emission
tomography: a feasibility study. Mol Imaging Biol. 2007;9:308–317.
8. Miyazawa H, Osmont A, Petit-Taboue MC, et al. Determination of18F-fluoro-2-
deoxy-D-glucose rate constants in the anesthetized baboon brain with dynamic
positron tomography. J Neurosci Methods. 1993;50:263–272.
9. Sokoloff L, Smith CB. Basic principles underlying radioisotopic methods for
assay of biochemical processes in vivo. In: Greitz T, Ingvar DH, Wid´ en L, eds.
The Metabolism of the Human Brain Studied with Positron Emission
Tomography. New York, NY: Raven Press; 1983:123–148.
10. Ohtake T, Kosaka N, Watanabe T, et al. Noninvasive method to obtain input
function for measuring glucose utilization of thoracic and abdominal organs.
J Nucl Med. 1991;32:1432–1438.
11. Dimitrakopoulou-Strauss A, Strauss LG, Mikolajczyk K, et al. On the fractal
nature of positron emission tomography (PET) studies. World J Nucl Med.
2003;2:306–313.
12. Benz MR, Allen-Auerbach MS, Eilber FC, et al. Combined assessment of
metabolic and volumetric changes for assessment of tumor response in patients
with soft-tissue sarcomas. J Nucl Med. 2008;49:1579–1584.
18F-FDG MONITORING IN SARCOMAS • Dimitrakopoulou-Strauss et al.557

Page 8

13. Evilevitch V, Weber WA, Tap WD, et al. Reduction of glucose metabolic activity
is more accurate than changes in size at predicting histopathologic response to
neoadjuvant therapy in high-grade soft-tissue sarcomas. Clin Cancer Res. 2008;
14:715–720.
14. Weber WA, Ott K, Becker K, et al. Prediction of response to preoperative
chemotherapy in adenocarcinomas of the esophagogastric junction by metabolic
imaging. J Clin Oncol. 2001;19:3058–3065.
15. Schuetze SM, Rubin BP, Vernon C, et al. Use of positron emission tomography
in localized extremity soft tissue sarcoma treated with neoadjuvant chemother-
apy. Cancer. 2005;103:339–348.
16. Tseng J, Dunnwald LK, Schubert EK, et al.
advanced breast cancer: correlation with tumor blood flow and changes in
response to neoadjuvant chemotherapy. J Nucl Med. 2004;45:1829–1837.
17. Hellwig D, Graeter TP, Ukena D, Georg T, Kirsch CM, Sch¨ afers HJ. Value of F-
18-fluorodeoxyglucose positron emission tomography after induction therapy of
18F-FDG kinetics in locally
locally advanced bronchogenic carcinoma. J Thorac Cardiovasc Surg. 2004;
128:892–899.
18. Cerfolio RJ, Bryant AS, Winokur TS, Ohja B, Bartolucci AA. Repeat FDG-PET
after neoadjuvant therapy is a predictor of pathologic response in patients with
non-small cell lung cancer. Ann Thorac Surg. 2004;78:1903–1909.
19. Dimitrakopoulou-Strauss A, Strauss LG, Burger C, et al. Prognostic aspects of
18F-FDG PET kinetics in patients with metastatic colorectal carcinoma receiving
FOLFOX chemotherapy. J Nucl Med. 2004;45:1480–1487.
20. Larson SM, Erdi Y, Akhurst T, et al. Tumor treatment response based on visual
and quantitative changes in global tumor glycolysis using PET-FDG imaging: the
visual response score and the change in total lesion glycolysis. Clin Positron
Imaging. 1999;2:159–161.
21. Boellaard R, Krak NC, Hoekstra OS, Lammertsma AA. Effects of noise, image
resolution, and ROI definition on the accuracy of standard uptake values: a
simulation study. J Nucl Med. 2004;45:1519–1527.
558THE JOURNAL OF NUCLEAR MEDICINE • Vol. 51 • No. 4 • April 2010