| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Clinical Trials |
Department of Radiology, Division of Nuclear Medicine [H. V., J. M. P., S. G. K.], Department of Pathology [R. A. S., M. L.], and Division of Thoracic Surgery [E. V., D. E. W.], University of Washington, Seattle, Washington 98195
 |
ABSTRACT |
|---|
 Top ABSTRACT INTRODUCTIONMATERIALS AND METHODSSurgical StagingPathologyStatistical AnalysisRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSSurgical StagingPathologyStatistical AnalysisRESULTSDISCUSSIONREFERENCES |
|---|
Tumor stage and histopathological grade are used to describe the extent
of disease and tumor aggressiveness. However, no comprehensive
pathological grading system exists for all NSCLCs. Tumor stage is the
strongest prognostic factor in NSCLC and the most important parameter
that guides treatment decisions. Overall 5-year survival rates by
pathological stage are: IA, 67%; IB, 57%; IIA, 55%; IIB, 38–39%;
IIIA, 23–25%; IIIB, 3–7%; IV, 1% (2)
. Surgery
represents the best chance for cure in NSCLC. However, surgery with
curative intent is an option in only 
30% of cases [stages I, II,
and selected IIIA
(T3N1M0)].
Even if a complete, presumably curative, resection can be performed,
>50% will relapse, with the majority of these relapses occurring at
distant sites (3, 4, 5)
. Hence, each pathological substage
remains a heterogeneous population containing individuals at
much higher risk of recurrence and death than others in the same
substage. Therefore, there is a need for a noninvasive grading system
to further characterize NSCLCs.
Measures of tumor proliferation have prognostic value in resected NSCLC (6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21) . However, these markers require a biopsy or resection and are subject to tissue sampling errors. FDG uptake has been shown to correlate with tumor proliferative rates in lymphomas and in head and neck cancers (22, 23, 24, 25, 26) . FDG uptake has also been shown to be a useful means of grading gliomas, head and neck cancers, lymphomas, breast cancer, and sarcomas (25 , 27, 28, 29, 30, 31, 32, 33, 34, 35) . Moreover, FDG uptake in focal pulmonary abnormalities has been shown to correlate with lesion doubling time (36) , a measure previously shown to predict lung tumor aggressiveness (37, 38, 39) . We therefore proposed to compare FDG uptake, quantitated as a SUV, with tumor proliferative rates assessed by immunohistochemistry (Ki-67 antigen score). Our hypothesis was that more metabolically active NSCLCs have higher proliferative rates (higher Ki-67 scores).
|
MATERIALS AND METHODS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSSurgical StagingPathologyStatistical AnalysisRESULTSDISCUSSIONREFERENCES |
|---|
FDG PET Imaging
Imaging.
All of the PET studies were performed on a General Electric Advance PET
tomograph. FDG imaging was performed in all of the patients to
quantitate FDG uptake as a proposed noninvasive measure of
primary-tumor aggressiveness (tumor grade) and to localize other
potential tumor sites as part of patient staging. Patients were asked
to fast for 12 h prior to tracer administration. Two i.v.
catheters were placed in opposite arms, one for tracer administration,
the other for blood sampling. An initial blood sample was obtained at
the time of i.v. catheter placement to screen for abnormally high
plasma glucose levels. Just prior to tracer administration, patients
also received 1 mg of i.v. lorazepam to decrease benign muscular uptake
in the neck and upper thorax. Seven to 11 mCi of FDG was infused i.v.
over a 2-min period using a Harvard pump (Harvard, Boston, MA). After a
45-min rest period, patients were placed supine in the scanner with the
thorax positioned to fit within two contiguous 15-cm wide tomograph
fields-of-view. Imaging always started with a 15-min-long emission scan
performed over the thoracic field-of-view encompassing the primary lung
cancer. This allowed us to quantitate tracer uptake (SUV) in the
primary tumor over a standard time period (45–60 min) after injection
to control for the time-dependence of the SUV. For each of 33 patients,
three blood samples were collected at 55 min after FDG injection to
measure plasma glucose and the glucose values averaged. The initial
blood glucose determination was used in the remaining six patients.
Plasma glucose levels were used to correct the FDG uptake values
provided by the General Electric Advance scanner. The other thoracic
field-of-view as well as the abdomen were also imaged with 10-min-long
emission scans. Five-min-long emission scans were performed over the
neck and pelvis. This was followed by 15-min-long transmission studies
over the three fields-of-view encompassing the chest and abdomen,
performed after all of the emission studies had been completed.
All of the studies were collected in two-dimensional imaging mode with scatter septae in place. Real-time randoms correction using counts obtained with a delayed coincidence window and deconvolution-based scatter corrections supplied by the manufacturer were applied. The raw PET data were reconstructed using the standard filtered back-projection available on the General Electric Advance PET system. The following reconstruction parameters were used: 12 mm Hanning filter, 55 cm image diameter, and 128 x 128 array size.
Quantitative Imaging Analysis.
The SUV was used to quantitate tumor FDG uptake. The SUV is defined as
the time-averaged tissue activity [C (µCi/ml)],
from 45 to 60 min after injection, divided by the injected dose
[ID (mCi)], per kg of patient body weight:
 |
All of the SUV data were corrected for partial volume effects, based on
the average diameter of the tumor (RCs). RCs have been estimated based
on lesion diameter from phantom measurements performed in the General
Electric Advance tomograph (40)
. These RCs have been
calculated for the image reconstruction parameters and filter used in
imaging of NSCLC patients (Fig. 1)
. These
RCs (RC) were applied to the difference between lesion and
background activity as defined by Eq. B. The maxSUV, instead of an
average SUV over a ROI, was used because it is less sensitive to
partial volume effects.
|
.
 |
Because the plasma glucose level affects the FDG SUV value, we
also evaluated the PV corr maxSUV scaled as:
 |
The BSA implementation of the SUV was also investigated for each tumor
(n = 39). The gluc-PV corr maxSUV was scaled to the BSA
as defined in Eq. D:
 |
 |
|
Surgical Staging |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSSurgical StagingPathologyStatistical AnalysisRESULTSDISCUSSIONREFERENCES |
|---|
The two patients that were not surgically staged consisted of:
(a) a patient who was found to be stage IV by FDG PET
imaging, with further confirmation of the metastases by magnetic
resonance imaging examination. This patient underwent
bronchoscopic biopsy of the primary tumor; and (b) a patient
who underwent a percutaneous biopsy but who was not a candidate for
resection because of poor medical condition (clinical stage IIIA by
PET). The distribution of patients among surgical stages is reported in
Table 1
.
|
|
Pathology |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSSurgical StagingPathologyStatistical AnalysisRESULTSDISCUSSIONREFERENCES |
|---|
|
Statistical Analysis |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSSurgical StagingPathologyStatistical AnalysisRESULTSDISCUSSIONREFERENCES |
|---|
|
RESULTS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSSurgical StagingPathologyStatistical AnalysisRESULTSDISCUSSIONREFERENCES |
|---|
: histological
subtype and tumor differentiation, surgical stage, tumor diameter,
Ki-67 score, maxSUV, PV corr maxSUV, and gluc-PV corr maxSUV. For
clarity, the values for the BSA definition of the SUV are not included.
The distribution of patients among surgical stages is: stage IA, 11;
stage IB, 15; stage IIA, 0; stage IIB, 4; stage IIIA, 3; stage IIIB, 3;
stage IV, 3 (Table 1)
.
For each tumor, the average lesion diameter was determined from CT
scans, and the corresponding RC was applied to the lesion maxSUV while
accounting for normal lung uptake (Eq. B). The correlation (Rho, 0.56;
P = 0.0006) between uncorrected maxSUV and lesion size
disappears (Rho, -0.035; P = 0.83) when applying the
RCs. The relationship between PV corr maxSUV values and lesion diameter
is shown in Fig. 2
. There is also no
correlation between maxSUV corrected for both partial volume and
glucose (gluc-PV corr maxSUV) and lesion size: Rho, -0.047;
P = 0.77.
|
. Further correction for plasma glucose
level (gluc-PV corr maxSUV) yielded a correlation with Ki-67 scores of:
Rho, 0.67; P < 0.0001. The body surface area
implementation of the SUV yielded the following correlation with Ki-67
scores for gluc-PV corr maxSUV-BSA: Rho, 0.645;
P < 0.0001.
|
The other implementations of SUV have the following correlations with Ki-67 for these stage I tumors: gluc-PV corr maxSUV: Rho, 0.715; P = 0.0003; and gluc-PV corr maxSUV-BSA: Rho, 0.74; P = 0.0002.
The best correlation between PV corr maxSUV and Ki-67 score was
found for the 15 patients with surgical stage IB
(T2N0
M0: Rho, 0.83; P = 0.0019; Fig. 4
). The other implementations of SUV have
the following correlations with Ki-67 for these stage IB tumors:
gluc-PV corr maxSUV: Rho, 0.76; P = 0.0047; and gluc-PV
corr maxSUV-BSA: Rho, 0.71; P = 0.0079. Within the
cluster of the five lesions with both very low FDG SUVs and low Ki-67
scores, four are bronchoalveolar carcinomas (Fig. 4)
. This histological
subtype of adenocarcinoma is known to have a better prognosis than
other NSCLCs.
|
).
|
).
|
|
DISCUSSION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSSurgical StagingPathologyStatistical AnalysisRESULTSDISCUSSIONREFERENCES |
|---|
Both static and dynamic imaging measures of tumor FDG uptake exist: SUV versus FDGMR. These methods vary greatly in their level of experimental complexity. We have elected to quantitate FDG uptake from static imaging data and with the SUV for the following reasons: (a) in order for FDG uptake quantitation to have widespread application as a means of grading tumors, it needs to be simple to measure without resorting to dynamic imaging, which would limit its clinical application; and (b) with both tumor uptake and whole-body staging information desired for NSCLC patients, the exclusion of dynamic imaging allows for shorter imaging protocols. This helps maximize patient enrollment and compliance with the imaging protocol.
The literature evidence and our experience indicate the SUV is an adequate and simple substitute for the experimentally complex FDGMR. The experience of our PET group in FDG imaging of sarcomas has shown a high degree of correlation between the FDGMR quantitated using dynamic imaging and Patlak analysis (44) and the maximum FDG SUV of those tumors over a very wide range of metabolic rates (45) . The correlation between FDGMR and FDG-SUV was better for tumors with high FDGMR values (greater than the mean of 11.7 µmol/min/100 g; r, 0.89) than for those with low FDGMR (r, 0.62). In addition, a strong correlation between SUV-lean and FDGMR (determined by dynamic imaging and graphical Patlak analysis) was found in 20 FDG PET studies performed on 10 patients with primary lung cancer (46) . Similar results were also reported in head and neck cancers and lymphomas (47) , and in breast cancer (48) . Nonetheless, we realize that the SUV may provide less information than the FDGMR and that the SUV also increases with time after tracer injection (49 , 50) . For this reason, every effort was made to standardize this measurement in time with each primary lung tumor being imaged for 15 min starting at 45 min after tracer injection.
Experience with lymphomas showed that high FDG uptake, measured as a high SUV, correlates with a high S-phase fraction (26) . Similarly, FDG SUV was found to correlate with the proliferative index measured by flow cytometry in head and neck tumors (23) . Recently, preliminary evidence from a series of 23 patients, presented at the 1998 Society of Nuclear Medicine meeting, points to a similar correlation of FDG SUV and proliferation (measured by PCNA) in NSCLC (51) . However, this study was limited mostly to lung adenocarcinomas (20 of 23 lesions), a subset of NSCLCs. This evidence suggests that the SUV represents an appropriate measure of tumor metabolism for correlation with specimen-derived markers of proliferation in NSCLC.
Tumor proliferative rates may be estimated in human tumor samples by mitotic figure counting, immunohistochemical detection of cell cycle-specific proteins (Ki-67 and PCNA), and DNA flow cytometry. Multiple studies provide evidence that these proliferation measures have prognostic significance in NSCLC.
A high mitotic index has been correlated with decreased survival in stage I NSCLC (6) , in T1N0 adenocarcinomas (7) , and in peripheral adenocarcinomas of less than 2-cm diameter (8) . Several flow cytometry studies of NSCLC have concluded that a high S-phase fraction was a negative prognostic factor for survival (9, 10, 11, 12) . Immunohistochemical staining for Ki-67 has also been identified as a predictor of survival in NSCLC (13, 14, 15, 16) , with lower disease-free interval in patients with highly proliferating tumors (17) . Increased PCNA expression was also found to predict poor outcome (18, 19, 20) and to be associated with the development of metastases (21) .
We elected to make correlations between tumor FDG uptake, and Ki-67 scores of NSCLC specimens. The selection of the Ki-67 marker was made in consideration of the following: (a) using mitotic figure counting, immunohistochemical techniques, and flow cytometry yield results that correlate with the results of the other methods. However, the correlation coefficients are only moderately strong because of inherent difficulties with each method; (b) mitotic figure counting is affected by interobserver variability, use of mitotic indices versus mitoses per area, time between tumor resection and fixation, and the laborious nature of the analysis; (c) expression of cell cycle-specific proteins, such as PCNA and the Ki-67-related antigen, is easier to assess semiquantitatively; and (d) DNA flow cytometry is complicated by technical difficulties in obtaining nuclei for analysis, by multiple cell populations in the tumor sample whose cell cycle compartments may be inextricably intermixed, and by various methods for calculating S-phase fractions (SPF).
The tumor pixel with the highest SUV was selected to represent the overall tumor grade as it corresponds to the most metabolically active region of the tumor. A representative section of the tumor was selected and stained for Ki-67 scoring. However, the stained section may not have exactly corresponded to the area of maximal FDG uptake. In addition, tumors may exhibit regional heterogeneity with respect to their Ki-67 score. Hence, this methodology may contribute to a less-than-perfect correlation between maxSUV and Ki-67 scores. However, exact registration of maxSUV pixel with the corresponding region of the resected mass is too combersome to be implemented.
The Ki-67 score reflects the percentage of tumor cells that are stained; the nontumoral cells are not included in the population so that this method of counting corrects for the stromal fraction. However, the FDG PET scan does not correct for this stromal fraction. For example, if the tissue voxel contains a high proportion/fraction of noncancerous cells, it will decrease the SUV of this voxel. The tissues with the lowest tumor fraction but the highest Ki-67 scores are, therefore, expected to give rise to the worst correlation between Ki-67 score and maxSUV. In addition, heterogeneity in tumor fraction within a given mass could worsen this correlation. This further discrepancy is minimized by: (a) selecting a representative tumor sample with the highest possible tumor fraction for Ki-67 staining and scoring; and (b) selecting the pixel with maximal SUV from the tumor volume at PET. This should favor the selection of the voxel with the highest tumor fraction.
Our finding of a correlation between FDG uptake and Ki-67 scores in NSCLC is compatible with the report by Duhaylongsod et al. (36) , in which FDG uptake in focal pulmonary abnormalities has been shown to correlate with lesion doubling time. Because the doubling time of a lesion reflects its rate of growth, the lesion proliferative rate would be expected to correlate with its FDG SUV. No statistically significant correlation was found in our series between the size and the Ki-67 score of a tumor. This should be expected because the size of a lesion is only a one-time measurement during its growth and does not reflect its growth rate as a proliferative score can.
The strong association present in our data between NSCLC FDG uptake and
the degree of tumor cell differentiation (Fig. 6)
is in keeping with
prior reported results. Higashi et al. (52)
reported a correlation between FDG uptake and degree of cell
differentiation in a series of lung adenocarcinomas, with
bronchoalveolar carcinomas having much lower uptake than
nonbronchioalveolar carcinomas. In a series of 22 squamous cell
carcinomas of the head and neck, Laubenbacher et al.
(53)
noted that higher FDG uptake was associated with
decreasing cell differentiation.
The finding of a correlation between NSCLC proliferation rate and FDG uptake has prognostic implications. Proliferation markers have prognostic significance in resectable NSCLC. Therefore, the correlation found between Ki-67 scores and FDG SUV suggests that FDG uptake can be used as a noninvasive measure of tumor grade and patient prognosis. The advantage of such a measure is that it is preoperative and would allow physicians to identify those patients with resectable NSCLC who have a worse prognosis. These patients could then be treated more aggressively with the administration of neo-adjuvant chemotherapy prior to resection. Patients with less aggressive tumors and a good prognosis would undergo resection only and would be spared the morbidity and cost of preoperative chemotherapy.
Because cell differentiation is one of the parameters used in all of
the tumor grading systems, the association we found between cell
differentiation and FDG uptake further supports the concept of
noninvasively grading NSCLCs with FDG PET. Within the cluster of the
five stage IB lesions with both very low FDG SUVs and low Ki-67 scores,
four were bronchoalveolar carcinomas (Fig. 4)
. This histological
subtype of adenocarcinoma is known to have a better prognosis than
other NSCLCs. This finding is compatible with the known prognostic
significance of cellular proliferation rates and with the hypothesis
that FDG uptake has prognostic value.
The most significant evidence of a relationship between NSCLC FDG uptake and prognosis is in the study by Ahuja et al. (54) . This retrospective study of FDG uptake in NSCLC showed that a SUV of >10 provided prognostic information independent of clinical stage and lesion size. However, in this study: (a) RCs were not used to correct for partial volume effects. Given the fact that corrections are necessary for lesions smaller than 2.8 cm imaged in the General Electric Advance scanner, the uptake of most T1 lesions (<3 cm) was likely underestimated; and (b) the start of imaging time, at least 30 min after injection, was not controlled. This can result in variations in SUV.
FDG is transported into tumor cells via glucose transporter membrane proteins (Glut1–5; Refs. 55 , 56 ). A recent study of 289 stage I NSCLCs demonstrated that increased expression of Glut1 and/or Glut3 is associated with poorer survival (57) . Although not yet shown in NSCLC, Glut1 immunoreactivity was found to correlate with Ki-67 cellular proliferation scores in a group of human breast tumors (58) . This evidence could provide a further link between the FDG uptake in stage I NSCLC and the prognosis and cellular proliferation.
Although the surgical stage is the most important prognostic factor to date, it provides an incomplete biological profile of NSCLC. The above results support a correlation between FDG uptake and proliferation, a known prognostic factor for NSCLC. FDG PET imaging can, therefore, provide preoperative prognostic information about the biological aggressiveness of these tumors. By helping to grade NSCLCs, FDG PET will contribute to the individualizing of patient therapy.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
1 This work was supported in part by NIH Grant
1R01 CA80907-01A1. 
2 To whom requests for reprints should be
addressed, at Department of Radiology, Division of Nuclear Medicine,
Box 356113, University of Washington Medical Center, 1959 North East
Pacific Street, Seattle, WA 98195. Phone: (206) 598-4240; Fax: (206)
598-4496; E-mail: vesselle{at}u.washington.edu
3 The abbreviations used are: NSCLC,
non-small cell lung cancer; PET, positron emission tomography; FDG,
[F-18]fluorodeoxyglucose; FDGMR, FDG metabolic rate; SUV,
standardized uptake value; maxSUV, (uncorrected lesion) maximum (pixel)
SUV; gluc-PV corr maxSUV, glucose- and partial volume-corrected
maxSUV; PV corr maxSUV, partial volume-corrected maxSUV; CT, computed
tomography; ROI, region-of-interest; BSA, body surface area; PCNA,
proliferating cell nuclear antigen; RC, recovery coefficient.
Received 5/19/00; accepted 7/ 6/00.
|
REFERENCES |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSSurgical StagingPathologyStatistical AnalysisRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  S. Bisdas, K. Spicer, and Z. Rumboldt Whole-Tumor Perfusion CT Parameters and Glucose Metabolism Measurements in Head and Neck Squamous Cell Carcinomas: A Pilot Study Using Combined Positron-Emission Tomography/CT Imaging AJNR Am. J. Neuroradiol., August 1, 2008; 29(7): 1376 - 1381. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. Kato, J. Shinoda, N. Nakayama, K. Miwa, A. Okumura, H. Yano, S. Yoshimura, T. Maruyama, Y. Muragaki, and T. Iwama Metabolic Assessment of Gliomas Using 11C-Methionine, [18F] Fluorodeoxyglucose, and 11C-Choline Positron-Emission Tomography AJNR Am. J. Neuroradiol., June 1, 2008; 29(6): 1176 - 1182. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 I. I. Na, B. H. Byun, H. J. Kang, G. J. Cheon, J. S. Koh, C. H. Kim, D. H. Choe, B.-Y. Ryoo, J. C. Lee, S. M. Lim, et al. 18F-Fluoro-2-Deoxy-Glucose Uptake Predicts Clinical Outcome in Patients with Gefitinib-Treated Non-Small Cell Lung Cancer Clin. Cancer Res., April 1, 2008; 14(7): 2036 - 2041. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. Kaira, N. Oriuchi, Y. Otani, K. Shimizu, S. Tanaka, H. Imai, N. Yanagitani, N. Sunaga, T. Hisada, T. Ishizuka, et al. Fluorine-18-{alpha}-Methyltyrosine Positron Emission Tomography for Diagnosis and Staging of Lung Cancer: A Clinicopathologic Study Clin. Cancer Res., November 1, 2007; 13(21): 6369 - 6378. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. C. Lee, J. L. Port, R. J. Korst, Y. Liss, D. N. Meherally, and N. K. Altorki Risk Factors for Occult Mediastinal Metastases in Clinical Stage I Non-Small Cell Lung Cancer Ann. Thorac. Surg., July 1, 2007; 84(1): 177 - 181. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. J. Downey, T. Akhurst, M. Gonen, B. Park, and V. Rusch Fluorine-18 fluorodeoxyglucose positron emission tomographic maximal standardized uptake value predicts survival independent of clinical but not pathologic TNM staging of resected non-small cell lung cancer J. Thorac. Cardiovasc. Surg., June 1, 2007; 133(6): 1419 - 1427. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Soret, S. L. Bacharach, and I. Buvat Partial-Volume Effect in PET Tumor Imaging J. Nucl. Med., June 1, 2007; 48(6): 932 - 945. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
W. A. Weber, J. Czernin, and M. E. Phelps Prognostic Significance of Fluorodeoxyglucose Uptake in Non-Small Cell Lung Cancer. A Blurry Picture? Clin. Cancer Res., June 1, 2007; 13(11): 3105 - 3106. [Full Text] [PDF]
|
|
|
|
|
|
H. Vesselle, J. D. Freeman, L. Wiens, J. Stern, H. Q. Nguyen, S. E. Hawes, P. Bastian, A. Salskov, E. Vallieres, and D. E. Wood Fluorodeoxyglucose Uptake of Primary Non-Small Cell Lung Cancer at Positron Emission Tomography: New Contrary Data on Prognostic Role Clin. Cancer Res., June 1, 2007; 13(11): 3255 - 3263. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. Hricak, P. L. Choyke, S. C. Eberhardt, S. A. Leibel, and P. T. Scardino Imaging Prostate Cancer: A Multidisciplinary Perspective Radiology, April 1, 2007; 243(1): 28 - 53. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. Schoder and M. Gonen Screening for Cancer with PET and PET/CT: Potential and Limitations J. Nucl. Med., January 1, 2007; 48(1_suppl): 4S - 18S. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K.-H. Lee, S.-H. Lee, D.-W. Kim, W. J. Kang, J.-K. Chung, S.-A. Im, T.-Y. Kim, Y. W. Kim, Y.-J. Bang, and D. S. Heo High fluorodeoxyglucose uptake on positron emission tomography in patients with advanced non-small cell lung cancer on platinum-based combination chemotherapy. Clin. Cancer Res., July 15, 2006; 12(14): 4232 - 4236. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
