Prediction model of chemotherapy response in osteosarcoma by 18F-FDG PET and MRI

Article (PDF Available)inJournal of Nuclear Medicine 50(9):1435-40 · August 2009with37 Reads
DOI: 10.2967/jnumed.109.063602 · Source: PubMed
Abstract
Response to neoadjuvant chemotherapy is a significant prognostic factor for osteosarcoma; however, this information can be determined only after surgical resection. If we could predict histologic response before surgery, it might be helpful for the planning of surgeries and tailoring of treatment. We evaluated the usefulness of (18)F-FDG PET for this purpose. A total of 70 consecutive patients with a high-grade osteosarcoma treated at our institute were prospectively enrolled. All patients underwent (18)F-FDG PET and MRI before and after neoadjuvant chemotherapy. We analyzed the predictive values of 5 parameters, namely, maximum standardized uptake values (SUVs), before and after (SUV2) chemotherapy, SUV change ratio, tumor volume change ratio, and metabolic volume change ratio (MVCR) in terms of their abilities to discriminate responders from nonresponders. Patients with an SUV2 of less than or equal to 2 showed a good histologic response, and patients with an SUV2 of greater than 5 showed a poor histologic response. The histologic response of a patient with an intermediate SUV2 (2 < SUV2 </= 5) was found to be predictable using MVCR. A patient with an MVCR of less than 0.65 is likely to be a good responder, whereas a patient with an MVCR of greater than or equal to 0.65 is likely to be a poor responder. According to our model, the predictive values for good responders and poor responders were 97% (31/32) and 95% (36/38), respectively. We found that combined information on (18)F-FDG PET and MRI scans, acquired before and after chemotherapy, could be used to predict histologic response to neoadjuvant chemotherapy in osteosarcoma.
Prediction Model of Chemotherapy Response
in Osteosarcoma by
18
F-FDG PET and MRI
Gi Jeong Cheon
1,2
, Min Suk Kim
3
, Jun Ah Lee
4
, Soo-Yong Lee
5
, Wan Hyeong Cho
5
, Won Seok Song
5
, Jae-Soo Koh
3
,
Ji Young Yoo
6
, Dong Hyun Oh
1
, Duk Seop Shin
7
, and Dae-Geun Jeon
5
1
Department of Nuclear Medicine, Korea Cancer Center Hospital, Seoul, Korea;
2
Department of Molecular Imaging Center, Korea
Institute of Radiological and Medical Sciences, Seoul, Korea;
3
Department of Pathology, Korea Cancer Center Hospital, Seoul,
Korea;
4
Department of Pediatrics, Korea Cancer Center Hospital, Seoul, Korea;
5
Department of Orthopedic Surgery, Korea Cancer
Center Hospital, Seoul, Korea;
6
Department of Radiology, Korea Cancer Center Hospital, Seoul, Korea; and
7
Department of
Orthopedic Surgery, Yeungnam University College of Medicine, Daegu, Korea
Response to neoadjuvant chemotherapy is a significant prog-
nostic factor for osteosarcoma; however, this information can
be determined only after surgical resection. If we could predict
histologic response before surgery, it might be helpful for the
planning of surgeries and tailoring of treatment. We evaluated
the usefulness of
18
F-FDG PET for this purpose. Methods: A
total of 70 consecutive patients with a high-grade osteosarcoma
treated at our institute were prospectively enrolled. All patients
underwent
18
F-FDG PET and MRI before and after neoadjuvant
chemotherapy. We analyzed the predictive values of 5 param-
eters, namely, maximum standardized uptake values (SUVs),
before and after (SUV2) chemotherapy, SUV change ratio, tumor
volume change ratio, and metabolic volume change ratio (MVCR)
in terms of their abilities to discriminate responders from nonre-
sponders. Results: Patients with an SUV2 of less than or equal to
2 showed a good histologic response, and patients with an SUV2
of greater than 5 showed a poor histologic response. The histo-
logic response of a patient with an intermediate SUV2 (2 , SUV2
# 5) was found to be predictable using MVCR. A patient with an
MVCR of less than 0.65 is likely to be a good responder, whereas
a patient with an MVCR of greater than or equal to 0.65 is likely
to be a poor responder. According to our model, the predic-
tive values for good responders and poor responders were
97% (31/32) and 95% (36/38), respectively. Conclusion: We
found that combined information on
18
F-FDG PET and MRI
scans, acquired before and after chemotherapy, could be used
to predict histologic response to neoadjuvant chemotherapy in
osteosarcoma.
Key Words: correlative imaging; oncology; PET;
18
F-FDG PET;
MRI; chemotherapy response; osteosarcoma
J Nucl Med 2009; 50:1435–1440
DOI: 10.2967/jnumed.109.063602
Tumor necrosis induced by neoadjuvant chemotherapy
has been reported to be the most powerful prognostic
indicator of survival in osteosarcoma patient s (1). More-
over, although histology is the accepted gold standard for
response evaluations, it is time- and labor-consuming and
is prone to inter- and intraobserver variabilities. Above all,
tumor necrosis rate can be determined only in resected
specimens, and thus, response monitoring during the course
of chemotherapy is not possible. To overcome these
limitations, other diverse imaging modalities have been
investigated (2,3). In particular, in osteosarcoma, these
modalities include bone scintigraphy (4), CT (5), MRI (6),
and
18
F-FDG PET (7).
Recently, PET using
18
F-FDG has been examined in the
context of determining prognosis (8), grading (9), staging
and restaging (10,11), guiding biopsy, and monitoring
response in many types of malignancies (12).
18
F-FDG
PET is a functional imaging modality and can detect
changes in tissue metabolism that usually precede structural
changes. Furthermore, in terms of response evaluations,
several reports have commented on the potential versatility
of
18
F-FDG PET (7,13–16).
In this prospective study, we evaluated the usefulness
of metabolic and volumetric information obtained by
18
F-FDG PET and MRI before and after the completion
of chem otherapy with the aim of predicting histologic
response to neoadjuvant chemotherapy in patients with
osteosarcoma.
MATERIALS AND METHODS
Patients
A total of 70 consecutive patients with osteosarcoma treated at
our institute were prospectively enrolled. Eligibility requirements
included primary high-grade osteosarcoma, the completion of
neoadjuvant chemotherapy and surgical resection, MRI and
18
F-
FDG PET scans obtained before and after neoadjuvant chemo-
therapy, a time between the first
18
F-FDG PET scan and the
initiation of chemotherapy of no more than 2 wk, and a time
between the second
18
F-FDG PET scan and surgery of no more
than 2 wk. Our institutional review board approved this study. All
patients provided written informed consent, and this study was
performed according to the ethical guidelines of our institutional
clinical research committee.
Received Mar. 1, 2009; revision accepted May 11, 2009.
For correspondence or reprints contact: Dae-Geun Jeon, Department of
Orthopedic Surgery, Korea Cancer Center Hospital, 215-4, Gongneung-
dong, Nowon-gu, Seoul, 139-706, Korea.
E-mail: dgjeon@kcch.re.kr.
COPYRIGHT ª 2009 by the Society of Nuclear Medicine, Inc.
OSTEOSARCOMA AND PET Cheon et al. 1435
Pretreatment Evaluation
Patients underwent a conventional evaluation (plain radiogra-
phy and MRI of the primary tumor, a
99m
Tc-methylene diphospho-
nate bone scan, and a CT scan of the chest) and
18
F-FDG PET
before neoadjuvant chemotherapy. Osteosarcoma diagnoses were
confirmed on the basis of histologic examinations of tumor tissues
obtained by open or needle biopsy, which was performed on
average 2.1 d (range, 1–5 d) before the first PET scan.
18
F-FDG PET/CT
18
F-FDG PET/CT scans were acquired using an integrated
PET/CT scanner (Discovery LS; GE Healthcare), which consisted
of a PET scanner (Advanced NXi; GE Healthcare) and an 8-slice
helical CT scanner (LightSpeed Plus; GE Healthcare). All patients
were instructed to fast for at least 6 h before the scans. Blood
glucose levels in all 70 patients were less than 6.6 mmol/L.
Truncal PET scans were obtained in 2-dimensional mode using
5–7 bed positions to ensure adequate coverage from head to pelvic
floor. Additional regional PET scans were also acquired in the
same manner as the truncal scans (using 3–5 bed positions) to
cover tumor sites located in the lower extremities. Emission scans
(5 min/frame; 128 · 128 matrix) were obtained 50 min after an
intravenous injection of
18
F-FDG (370 MBq). In the case of
children (under 15 y), 7.4 MBq of
18
F-FDG per kilogram of body
weight (mean, 222 MBq; range, 185–333 MBq) were injected
intravenously. CT scans were obtained immediately before PET
scans, using a multidetector helical CT scanner. The imaging
parameters used were as follows: 140 kVp, 80 mA, 0.8 s/CT
rotation, pitch of 6, and a 22.5 mm/s table speed. No contrast
material was administered. CT images were created using a 512 ·
512 matrix but were reduced to a 128 · 128 matrix to correspond
to PET emission images. PET/CT images were reconstructed
using CT scans for attenuation correction and the ordered-subset
expectation maximization algorithm (2 iterations, 16 subsets),
as previously described (17). Images were coregistered using
dedicated software (eNTEGRA; GE Healthcare).
18
F-FDG PET/CT Image Interpretation
Abnormal
18
F-FDG uptake was defined as uptake greater than
background uptake in surrounding tissues that did not exhibit
tracer uptake. Areas of abnormal
18
F-FDG uptake were identified,
and intensities of
18
F-FDG uptake were quantified by calculating
standardized uptake values (SUVs) from amounts of
18
F-FDG
injected, total body weight, and regional uptake in attenuation-
corrected regional images. Specifically, SUV was defined as
maximum SUV (SUVmax) of the region of interest (ROI) and
calculated by the following equation: (activity/unit volume)/
(injected dose/total body weight). All PET/CT scans were reviewed
and interpreted by an experienced nuclear physician.
MRI
MRI sequences included a standard (spin-echo) T1-weighted
sequence (repetition time [ms]/echo time [ms], 400–900/10–20),
with or without gadolinium enhancement, and an intermediate-
weighted/T2-weighted sequence (1,500–2,500/70–100), without fat
suppression. Intramedullary tumor lengths were measured in cor-
onal sections of unenhanced T1-weighted sequences, and tumor
widths and depths were measured in axial enhanced T1- and T2-
weighted sequences without fat suppression (18). MR images were
independently reviewed by 2 of the authors of this article. When the
2 reviewers found a size discrepancy of more than 10%, images were
reviewed simultaneously, and decisions were made by consensus.
Neoadjuvant Chemotherapy
All patients underwent 2 cycles of preoperative chemotherapy
using the modified T10 protocol (19). Briefly, each cycle of
chemotherapy consisted of high-dose methotrexate, adriamycin,
and cisplatin. Methotrexate was administered twice at a dose of
8–12 g/m
2
on days 1 and 7. Cisplatin was administered at a dose
of 100 mg/m
2
on day 14 over 2 h. Subsequently, adriamycin was
TABLE 1. Patient Characteristics
Characteristic Value
Age (n)
#15 y 41 (58.6%)
.15 and #40 y 25 (35.7%)
.40 y 4 (5.7%)
Sex (n)
Male 48 (68.6%)
Female 22 (31.4%)
AJCC stage (n)
IIA 24 (34.3%)
IIB 40 (57.1%)
IV 6 (8.6%)
Tumor volume (cm
3
)
Median 149
Range 17–2,882
Location (n)
Distal femur 35 (50.0%)
Proximal tibia 17 (24.3%)
Proximal humerus 6 (8.6%)
Others 12 (17.1%)
Pattern on plain radiograph (n)
Lytic 17 (24.3%)
Blastic 26 (37.1%)
Mixed 27 (38.6%)
Pattern on MRI (n)
Concentric 60 (85.7%)
Longitudinal 10 (14.3%)
SUV1
Median 8.0
Range 2.4–47.5
SUV2
Median 4.5
Range 1.5–16.6
Time from first PET to initiation of chemotherapy
Median 6 d
Range 1–13 d
Time from end of chemotherapy to second PET
Median 19 d
Range 16–22 d
Time from second PET to surgery
Median 2 d
Range 1–13 d
Pathologic subtype (n)
Osteoblastic 50 (71.4%)
Chondroblastic 13 (18.6%)
Fibroblastic 5 (7.1%)
Other 2 (2.9%)
Type of surgery (n)
Amputation 3 (4.3%)
Limb salvage 67 (95.7%)
Histologic response (n)
Good 33 (47.1%)
Poor 37 (52.9%)
AJCC 5 American Joint Committee on Cancer.
1436 THE JOURNAL OF NUCLEAR MEDICINE Vol. 50 No. 9 September 2009
administered at 60 mg/m
2
over 18 h. The intervals between the
end of the first cycle of chemotherapy and initiation of the second
cycle, and between the end of the second cycle of chemotherapy
and surgery, were around 3 wk.
Histologic Assessments of Response to
Preoperative Chemotherapy
The effects of preoperative chemotherapy were graded histo-
logically as described by Rosen et al.: grades III and IV (.90%
necrosis) indicate good response and grades I and II (,90%
necrosis) indicate poor response (19).
Definitions and Calculations of Parameters
We defined prechemotherapy SUVmax as SUV1 and preoper-
ative SUVmax as SUV2. Tumor volume (TV) was calculated
using the ellipsoid formula, as described previously (18). The
following parameters were calculated using these values:
SUV change ratio ðSCRÞ 5 SUV2=SUV1
Volume change ratio ðVCRÞ 5 TV after chemotherapy=
TV before chemotherapy
Metabolic volume change ratio ðMVCRÞ 5 SCR · VCR
Statistics
We analyzed the predictive values of 5 parameters—SUV1,
SUV2, SCR, VCR, and MVCR—in terms of their abilities to
discriminate responders from nonresponders. For this purpose, we
used receiver-operating-characteristic (ROC) curves and calcu-
lated areas under curves (AUCs) for each parameter. We chose
parameters that best predicted response and determined cutoffs
that showed highest accuracy. Then we grouped the patients using
these cutoffs and calculated positive and negative predictive
values. Finally, we devised a decision tree for predicting response
based on imaging parameters and calculated predictive values
using this model. All calculations were performed using SPSS
(version 13.0; SPSS Inc.). All P values were derived from the
2-sided test, and values of less than 0.05 were considered significant.
RESULTS
Patients’ Characteristics
Patients’ characteristics are detailed in Table 1. The
median age of patients was 14 y (range, 5–59 y), and 69%
of patients were male. The median tumor volume was 149
cm
3
(range, 17–2,882 cm
3
). On the basis of the revised
American Joint Committee on Cancer staging system, 24
patients (34.3%) had a stage IIA tumor, 40 (57.1%) had a
stage IIB tumor, and 6 (8.6%) had a stage IV tumor. Half of
the 70 patients presented with a tumor in the distal femur.
The median tumor SUV1 and SUV2 values were 8.0 and
4.5, respectively. Median time between the first PET
examination and the initiation of chemotherapy was 6 d.
FIGURE 1. ROC curve analysis of response prediction. ROC curves of SUV1 (A), SUV2 (B), SCR (C), VCR (D), and MVCR (E)
were plotted to predict histologic response. On basis of AUC, all parametersexcept SUV1predicted histologic response.
OSTEOSARCOMA AND PET Cheon et al. 1437
Median time between the end of chemotherapy and the
second PET examination was 19 d (range, 16–22 d), and
median time between the second PET examination and
surgery was 2 d. Thirty-three patients (47.1%) showed good
histologic response in resected specimens after neoadjuvant
chemotherapy.
ROC Curve Analysis of Response Prediction
Before plotting the ROC curves, we checked the bivar-
iate association between histologic response and clinico-
pathologic parameters; however, no significa nt correlation
was found. ROC curves of SUV1, SUV2, SCR, VCR, and
MVCR were plotted to predict histologic response (Fig. 1).
All parameters, except SUV1, predicted histologic re-
sponse. We calculated the predictive values of SUV2,
SCR, VCR, and MVCR at each cutoff value (Table 2).
We found that low (#2) and high (.5) SUV2 values
predicted histologic response and, therefore, set 2 cutoff
values for SUV2. On the basis of these cutoffs, we divided
patients into 3 groups: group I (SUV2 # 2; n 5 7), group II
(2 , SUV2 # 5; n 5 33), and group III (SUV2 . 5; n 5
30). All 7 patients in group I were good responders,
whereas most of the patients (28/30) in group III were
poor responders. Of the 33 patients in group II, 24 (73%)
were good responders and 9 (27%) were poor responders.
ROC Curve Analysis for Group II Patients
To predict histologic response for group II patients, we
drew ROC curves and calculated the AUCs of SUV1, SCR,
VCR, and MVCR. MVCR was found to predict histologi c
response best among group II patients (AUC, 0.93). All 8
patients with an MVCR of greater than or equal to 0.65
showed a poor response, whereas most patients (24/25)
with an MVCR of less than 0.65 showed a good response.
Decision Tree for Response Prediction
A decision tree was devised to predict histologic re-
sponse based on SUV2 and MVCR values (Fig. 2).
According to our model, 32 patients were predicted to be
good responders and the remaining 38 to be poor re-
sponders. Predictive values for good responders and poor
responders were 97% (31/32) and 95% (36/38), respec-
tively. Briefly, patients with an SUV2 of less than or equal
to 2 showed a good histologic response (predictive value,
100%; 95% confidence interval [CI] , 77%2100%), and
patients with an SUV2 of greater than 5 showed a poor
histologic response (predictive value, 93%; 95% CI,
87%293%). The histologic response of a patient with an
intermediate SUV2 (2 , SUV2 # 5) was found to be
predictable using MVCR; for example, a patient with an
MVCR of less than 0.65 is likely to be a good responder
(predictive value, 96%; 95% CI, 89%296%), whereas a
patient with an MVCR of greater than or equal to 0.65 is
likely to be a poor responder (predictive value, 100%; 95%
CI, 77%2100%).
DISCUSSION
Unlike morphologic imaging modalities,
18
F-FDG PET
reflects the metabolic rate of glycolysis in tumors, and thus,
18
F-FDG PET should be more accurate for assessing
treatment response because it can more correctly identify
viable residual tumors. Most studies conducted on this topic
have demonstrated a strong correlation between a reduction
in tumor glucose metabolism after chemotherapy and tumor
necrosis rate. Th ese studies used various types of
18
F-FDG
uptake indices—such as tumor-to-background ratio (TBR)
(9,13,16), SUV2 (14), and SUV2-to-SUV1 ratios (15)—and
have suggested various cutoff values for
18
F-FDG uptake
indices to predict response (Table 3). However, mos t of
TABLE 2. Predictive Values of SUV2, SCR, VCR, and
MVCR in 70 Patients
Parameter
Cutoff
value n GR/PR PPV (%) NPV (%)
Accuracy
(%)
SUV2 #2 7 7/0 100 59 63
#3 19 15/4 79 65 69
#4 29 23/6 79 76 77
#5 40 31/9 78 93 84
#6 46 31/15 67 92 76
#7 52 33/19 63 100 73
SCR #0.3 8 7/1 88 58 61
#0.4 16 14/2 88 65 70
#0.5 26 21/5 81 73 76
#0.6 35 26/9 74 80 77
#0.7 47 30/17 64 87 71
#0.8 50 30/20 60 85 67
VCR #0.8 16 10/6 63 57 59
#1.0 29 22/7 76 73 74
#1.2 46 28/18 61 79 67
MVCR #0.2 8 7/1 88 58 61
#0.4 19 16/3 84 67 71
#0.6 34 28/6 82 86 84
#0.8 47 32/15 68 96 77
#1.0 52 32/20 62 94 70
GR 5 good responder; PR 5 poor responder; PPV 5 positive
predictive value; NPV 5 negative predictive value; SCR 5 SUV2/
SUV1; VCR 5 tumor volume after chemotherapy/tumor volume
before chemotherapy; MVCR 5 SCR · VCR.
FIGURE 2. Decision tree for response prediction. Decision
tree was devised to predict histologic response based on
SUV2 and MVCR values. Predictive values of our model for
good responders and poor responders were 97% (31/32)
and 95% (36/38), respectively.
1438 THE JOURNAL OF NUCLEAR MEDICINE Vol. 50 No. 9 September 2009
these studies lacked statistical power because of low patient
numbers.
Pr evious studies have used tw o
18
F-FDG PET scan
parameters, namely, TBR and SUVmax. SUVmax r epre-
sents the highest metabolic activity point in a tumor,
whereas the TBR (or mean SUV) represents mea n activity
in the ROI. In sarcoma patients, TBR values, which are
highly dependent on ROIs analyzed, are known to be more
prone to interobserver variability than SUV max (20). In 3
studies that used TBR (7,13,16), TBR was calculated by
dividing average tumor uptakes in ROIs by average uptake
in contralateral ROIs, and in these studies 3 different
cutoffs of TBR-change ratios before and after chemother-
apy were used. Although it is difficult to compare TBR
directly with SUVmax because of indivi dual differences
in background uptake and body weight, changes in their
ratios (TBRpost/TBRpre and SU Vpost/SUVpre) are likely
to be comparable. In the present study, when we applied
these 3 cutoffs to our patients (TBRpost/TBRpre, 0.46,
0.6, and 0.7), predictive values for good responders were
85%, 74%, a nd 64%, respectively, a nd those for po or
responders were 70%, 80%, and 87%, respe ctively (Table
2). In 2 previous studies, SUVmax was used a parameter
for response predi ction (14 , 1 5). In these studies, it was
found that SUV2/SUV1 and SUV2 correlated with histol-
ogic response and that an SUV2 of less than 2 and an
SUV2/SUV1 of less than 0.4 were good response indica-
tors. These results are partly concordant with ours, but it is
difficult to draw a definite conclusion from the data
presented because these previous studies included fewe r
than 20 osteosarcoma patients and more than 70% were
poor responders.
In the present study, we used SUVmax—which repre-
sents the highest metabolic activity—as a semiquantitative
parameter of
18
F-FDG uptake. Our data show that SUV2
rather than SUV1 can predict response, suggesting that
18
F-FDG PET after chemotherapy can identify residual
tumors m ore correctly than that before chemotherapy and,
thus, predict histologic response more accurately. Further-
more, SUV2 was found to be an independent predictor
of response to chemotherapy when it was high (.5) or
low (#2), regardless of residual tumor volumetric informa-
tion. However, SUV2 alone could not precisely predict
histologic response for intermediate values of SUV2 (2–5).
For tumors with intermed iate SUV2, MVCRs—which
encompass both SUV and tumor volume changes—were
found to have predictive value. Furthermore, our model was
able to predict histologic response correctly in 67 of 70
patients; its predictive values for good and poor responders
were 97% and 95%, respectively.
Currently, delineations of tumor margins and the mea-
surements of tumor volumes on PET scans are problematic,
because of the arbitrariness of tumor margin cutoff values.
Therefore, direct measurement of metabolic volume on a
PET scan is difficult. Tumor volume measurements are
easier and more reproducible in osteosarcoma than in other
tumor types, when an appropriate MRI technique is used
(21). Furthermore, in the present study, we used a new
parameter, MVCR, to reflect both metabolic and volumetric
changes induced by chemotherapy. For all 70 patients,
MVCR could discriminate good responders from poor
responders with a predictive value of 82% (31/38) and
94% (30/32), respectively (cutoff of MVCR, 0.65). How-
ever, in those cases with an intermediate SUV2, MVCR
could correctly predict histologic response. Thus, the
approach we suggest uses SUV2 initially to predict re-
sponse and follows this with MVCR in those with an
intermediate SUV 2 value.
Nevertheless, the present study ha s some inhere nt lim-
itations. First, it is a singl e-center study with a relatively
TABLE 3. Summary of Previous Studies
P1 P2
Reference n* GR/PR P Mean Median Range Mean Median Range Cutoff
y
PPV NPV
Schulte
et al. (7)
27 17/10 TBR 10.3 3.3–33.2 2.92 0.32–20.33 P2/P1 # 0.6
for GR (n 5 19)
89.5% 100%
Franzius
et al. (13)
11 9/2 TBR 4.4 2.3–13.6 1.7 0.9–11.9 P2/P1 , 0.7
for GR (n 5 9)
100% 100%
Ye et al.
(16)
15 8/7 TBR 7.1 3.0–20.6 3.1 1.1–6.5 P2/P1 , 0.46
for GR (n 5 8)
100% 100%
Hawkins
et at. (14)
18 5/13 SUVmax 8.2 2.5–24.1 3.3 1.6–12.8 P2 , 2 for
GR (n 5 4)
75% 85.7%
Huang
et al. (15)
10 2/8 SUVmax 8.2 1.4–13.6 4.4 1.7–9.6 P2/P1 , 0.4
for GR (n 5 2)
100% 100%
Present
study
70 33/37 SUVmax 8.0 2.4–47.5 4.5 1.5–16.6 Algorithm for
GR (n 5 32)
97% 95%
*Patients with high-grade osteosarcoma were included in this table.
y
n 5 number of patients who met cutoff criteria.
GR 5 good responder; PR 5 poor responder; P 5 parameter; P1 5 parameter before chemotherapy; P2 5 parameter after
chemotherapy; PPV 5 positive predictive value; NPV 5 negative predictive value; TBR 5 tumor-to-background ratio.
OSTEOSARCOMA AND PET Cheon et al. 1439
small number of patients. Therefore, the prediction model
derived must be val idated prospectively in a larger p atient
population. Second, the partial-volume effect produced by
limited spatial resolution may have caused
18
F-FDG up-
take underest imations in sm all or necrotic lesions, which
would reduce SUV accuracy. Third, VCRs calculated
using the ellipsoidal formula might have over- or under-
estimated tumor volume. Fourth, we did not a nalyz e the
cost-effectiveness of
18
F-FDG PET. Fifth, our suggested
SUV cuto ffs may differ for different
18
F-FDG PET scan-
ners. Finally, we did not use mor e a quantita tive fo rm of
analysis,suchasPatlakgraphical analysis or other kinetic
methods.
CONCLUSION
We found that
18
F-FDG PET provides a useful tool for
predicting histologic response in osteosarcoma patients. We
suggest that combined metabolic and volumetric informa-
tion on
18
F-FDG PET and MRI scans, acquired before and
after completing chemotherapy, could be used to predict
histologic response to neoadjuvant chemotherapy in osteo-
sarcoma. A future external validation of our prediction
model is mandatory. We hope that this study proves to be of
benefit during surgical planning, but we emphasize that
before using our model to assess response to chemot herapy
or tailor neoadjuvant treatment, further confirmatory stud-
ies are required.
ACKNOWLEDGMENT
This work was partly supported by Korea Science and
Engineering Foundation (KOSEF) grant M20702010002-
08N0201-00000, funded by the Korean government (MEST).
REFERENCES
1. Davis AM, Bell RS, Goodwin PJ. Prognost ic factors in osteosarcoma: a critical
review. J Clin Oncol. 1994;12:423–431.
2. van der Woude HJ, Bloem JL, Hogendoorn PC. Preoperative evaluation and
monitoring chemotherapy in patients with high-grade osteogenic and Ewing’s
sarcoma: review of current imaging modalities. Skeletal Radiol. 1998;27:57–71.
3. Brisse H, Ollivier L, Edeline V, et al. Imaging of malignant tumours of the long
bones in children: monitoring response to neoadjuvant chemotherapy and
preoperative assessment. Pediatr Radiol. 2004;34:595–605.
4. Soderlund V, Larsson SA, Bauer HC, Brosjo O, Larsson O, Jacobsson H. Use of
99m
Tc-MIBI scintigraphy in the evaluation of the response of osteosarcoma to
chemotherapy. Eur J Nucl Med. 1997;24:511–515.
5. Wellings RM, Davies AM, Pynsent PB, Carter SR, Grimer RJ. The value of
computed tomographic measurements in osteosarcoma as a predictor of response
to adjuvant chemotherapy. Clin Radiol. 1994;49:19–23.
6. Holscher HC, Bloem JL, van der Woude HJ, et al. Can MRI predict the
histopathological response in patients with osteosarcoma after the first cycle of
chemotherapy? Clin Radiol. 1995;50:384–390.
7. Schulte M, Brecht-Krauss D, Werner M, et al. Evaluation of neoadjuvant
therapy response of osteogenic sarcoma using FDG PET. J Nucl Med. 1999;40:
1637–1643.
8. Franzius C, Bielack S, Flege S, Sciuk J, Jurgens H, Schober O. Prognostic
significance of
18
F-FDG and
99m
Tc-methylene diphosphonate uptake in primary
osteosarcoma. J Nucl Med. 2002;43:1012–1017.
9. Adler LP, Blair HF, Makley JT, et al. Noninvasive grading of musculoskeletal
tumors using PET. J Nucl Med. 1991;32:1508–1512.
10. Franzius C, Daldrup-Link HE, Wagner-Bohn A, et al. FDG-PET for detection of
recurrences from malignant primary bone tumors: comparison with conventional
imaging. Ann Oncol. 2002;13:157–160.
11. Kleis M, Daldrup-Link H, Matthay K, et al. Diagnostic value of PET/CT for the
staging and restaging of pediatric tumors. Eur J Nucl Med Mol Imaging.
2009;36:23–36.
12. Brenner W, Bohuslavizki KH, Eary JF. PET imaging of osteosarcoma. J Nucl
Med. 2003;44:930–942.
13. Franzius C, Sciuk J, Brinkschmidt C, Jurgens H, Schober O. Evaluation of
chemotherapy response in primary bone tumors with F-18 FDG positron
emission tomography compared with histologically assessed tumor necrosis.
Clin Nucl Med. 2000;25:874–881.
14. Hawkins DS, Rajendran JG, Conrad EU III, Bruckner JD, Eary JF. Evaluation
of chemotherapy response in pediatric bone sarcomas by [F -18]-fluorodeoxy-
D-glucose positron emission tomography. Cancer. 2002;94:3277–3284.
15. Huang TL, Liu RS, Chen TH, Chen WY, Hsu HC, Hsu YC. Comparison between
F-18-FDG positron emi ssion tomography and histology for the assessment of
tumor necrosis rates in primary osteosarcoma. J Chin Med Assoc. 2006;69:372–
376.
16. Ye Z, Zhu J, Tian M, et al. Response of osteogenic sarcoma to neoadjuvant
therapy: evaluated by
18
F-FDG-PET. Ann Nucl Med. 2008;22:475–480.
17. Na II, Cheon GJ, Choe DH, et al. Clinical significance of
18
F-FDG uptake by N2
lymph nodes in patients with resected stage IIIA N2 non-small-cell lung cancer:
a retrospective study. Lung Cancer. 2008;60:69–74.
18. Kim MS, Lee SY, Cho WH, et al. Tumor necrosis rate adjusted by tumor volume
change is a better predictor of survival of localized osteosarcoma patients. Ann
Surg Oncol. 2008;15:906–914.
19. Rosen G, Marcove RC, Huvos AG, et al. Primary osteogenic sarcoma: eight-year
experience with adjuvant chemotherapy. J Cancer Res Clin Oncol. 1983;
106(suppl):55–67.
20. Benz MR, Evilevitch V, Allen-Auerbach MS, et al. Treatment monitoring by
18
F-
FDG PET/CT in patients with sarcomas: interobserver variability of quantitative
parameters in treatment-induced changes in histopathologically responding and
nonresponding tumors. J Nucl Med. 2008;49:1038–1046.
21. Saifuddin A. The accuracy of imaging in the local staging of appendicular
osteosarcoma. Skeletal Radiol. 2002;31:191–201.
1440 THE JOURNAL OF NUCLEAR MEDICINE Vol. 50 No. 9 September 2009
    • "18 F-FDG PET/CT is now widely used in the initial diagnosis , staging and detection of recurrence in many kinds of can- cer [9][10][11][12][13][14] . The role of 18 F-FDG PET/CT in predicting response to chemotherapy in bone sarcomas [15][16][17][18][19][20][21][22] and softtissue sarcomas [23][24][25]has been assessed in many studies with contradictory results [19, 26] . However, histological het- erogeneity [23, 24], limited numbers of patients included [16, 19], and especially the lack of uniform treatment [15] make the interpretation of results difficult. "
    [Show abstract] [Hide abstract] ABSTRACT: PurposeThe histological response to neoadjuvant chemotherapy is an important prognostic factor in patients with osteosarcoma (OS) and Ewing sarcoma (EWS). The aim of this study was to assess baseline primary tumour FDG uptake on PET/CT, and serum values of alkaline phosphatase (ALP) and lactate dehydrogenase (LDH), to establish whether these factors are correlated with tumour necrosis and prognosis. Methods Patients treated between 2009 and 2014 for localized EWS and OS, who underwent FDG PET/CT as part of their staging work-up, were included. The relationships between primary tumour SUVmax at baseline (SUV1), SUVmax after induction chemotherapy (SUV2), metabolic response calculated as [(SUV1 − SUV2)/SUV1)] × 100, LDH and ALP and tumour response/survival were analysed. A good response (GR) was defined as tumour necrosis >90 % in patients with OS, and grade II-III Picci necrosis (persitence of microscopic foci only or no viable tumor) in patients with Ewing sarcoma. ResultsThe study included 77 patients, 45 with EWS and 32 with OS. A good histological response was achieved in 53 % of EWS patients, and 41 % of OS patients. The 3-year event-free survival (EFS) was 57 % in EWS patients and 48 % OS patients. The median SUV1 was 5.6 (range 0 – 17) in EWS patients and 7.9 (range 0 – 24) in OS patients (p = 0.006). In EWS patients the GR rate was 30 % in those with a high SUV1 (≥6) and 72 % in those with a lower SUV1 (p = 0.0004), and in OS patients the GR rate was 29 % in those with SUV1 ≥6 and 64 % in those with a lower SUV1 (p = 0.05). In the univariate analysis the 3-year EFS was significantly better in patients with a low ALP level (59 %) than in those with a high ALP level (22 %, p = 0.02) and in patients with a low LDH level (62 %) than in those with a high LDH level (37 %, p = 0.004). In EWS patients the 3-year EFS was 37 % in those with a high SUV1 and 75 % in those with a low SUV1 (p = 0.004), and in OS patients the 3-year EFS was 32 % in those with a high SUV1 and 66 % in those with a low SUV1 (p = 0.1). Histology, age and gender were not associated with survival. In the multivariate analysis, SUV1 was the only independent pretreatment prognostic factor to retain statistical significance (p = 0.017). SUV2 was assessed in 25 EWS patients: the median SUV2 was 1.9 (range 1 – 8). The GR rate was 20 % in patients with a high SUV2, and 67 % in those with a low SUV2 (p = 0.02). A good metabolic response (SUV reduction of ≥55 %) was associated with a 3-year EFS of 80 % and a poor metabolic response with a 3-year EFS of 20 % (p = 0.05). In the OS patients the median SUV2 was 2.7 (range 0 – 4.5). Neither SUV2 nor the metabolic response was associated with outcome in OS patients. ConclusionFDG PET/CT is a useful and noninvasive tool for identifying patients who are more likely to be resistant to chemotherapy. If this finding is confirmed in a larger series, SUV1, SUV2 and metabolic response could be proposed as factors for stratifying EWS patients to identify those with high-grade localized bone EWS who would benefit from risk-adapted induction chemotherapy.
    Full-text · Article · Sep 2016
    • "Pre-operative diagnostic imaging usually evaluates tumors with regard to morphology, size, location, extent, and presence of secondary complications, and determines the regional or distant metastasis. Positron emission tomography-computed tomography (PET-CT) is considered to be one of the most sensitive diagnostic modalities for evaluating tumor stage, response to therapy, and recurrence [1,3,6] . 18 Ffluorodeoxyglucose ( 18 F-FDG) uptake provides useful information during follow-up PET-CT examination; specifically, if metabolic activity significantly decreases in a malignant lesion, it can be considered a good response to therapy. "
    [Show abstract] [Hide abstract] ABSTRACT: This report describes the usefulness of positron emission tomography- computed tomography 1 (PET-CT) in evaluating recurrent or residual tumor following surgery. CT and (18)F-2 fluorodeoxyglucose PET-CT were pre- and post-operatively applied to multiple masses in a 3 dog with hemangiosarcoma. The distinction between the left subcutaneous mass and the 4 peritoneum was clarified on pre-operative CT examination, and malignancy was suspected 5 based on PET-CT. A recurrent or residual tumor in the left subcutaneous region was suspected 6 on post-operative PET-CT, and confirmed through histopathologic examination.
    Full-text · Article · Dec 2015
    • "Kern et al (11) first applied FDG-PET to soft tissue tumors, including malignant fibrous histiocytoma; it has since been shown to be one of the most powerful diagnostic tools in oncology, enabling the functional assessment of soft tissue tumors. Currently, FDG-PET can identify the metabolic rate of glycolysis in tumors and is increasingly applied to grading (12,13), staging (14), chemotherapeutic response assessment (15,16) and surgical planning (3) of soft tissue tumors. Preliminary reports emphasized the ability of FDG-PET to distinguish benign from malignant tumors (1–3,17). "
    [Show abstract] [Hide abstract] ABSTRACT: The aim of the current study was to evaluate the limitations of 2-deoxy-2-F(18)-fluoro-D-glucose positron emission tomography combined with computed tomography (FDG-PET/CT) when monitoring soft tissue tumors. The diagnostic criteria of malignancy was defined as the tumor having a maximum standardized uptake value (SUVmax) ≥2.0 and a maximum diameter ≥5 cm as measured using FDG-PET/CT. One-hundred-and-thirteen patients, that were either included in the criteria or not, were compared. In addition, the values of SUVmax of the primary tumor and relapse in 12 patients were evaluated. The Kaplan-Meier analysis demonstrated that patients with tumors measuring ≥5 cm size and ≥2.0 SUVmax were associated with a worse survival rate. Among the 12 patients with relapse, statistical significances were detected in the tumor diameters, however, not in the SUVmax values. Thus, the criteria identified patients that were associated with a poor prognosis, and the SUVmax of distant metastases and local recurrences were identified to be significantly affected by tumor size.
    Full-text · Article · Apr 2014
Show more