| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Clinical Investigations |
Department of Cancer Medicine [P. W., P. P.], MRC Cyclotron Unit [R. N. G., C. S., A. S. R., F. B., S. O., T. J.], and Department of Histopathology [M. A., M. G.], Imperial College School of Medicine, Hammersmith Hospital, London W12 0HS, United Kingdom
 |
ABSTRACT |
|---|
 Top ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
There are a number of important issues relevant to determining the optimal PET tracer for imaging tumor proliferation. The interpretation of PET data can be complicated by the presence of labeled metabolites of the injected tracer. The position of the label on the tracer will influence the type and number of labeled metabolites, and the isotope half-life influences the quality of the PET data. The presence of a high concentration of labeled metabolites may mask the signal of the parent compound. Pyrimidine nucleoside incorporation into DNA has been used to study proliferation for many years. Thymidine and its analogues, such as the halogenated pyrimidines BrdUrd and IUdR, are particularly useful because of their specificity for DNA. Although the long half-life of [124I]IUdR is attractive for protracted studies over several days, its use may be limited because of low tumor radioactivity compared with high levels of tissue metabolites (5) . Similarly [76Br]BrdUrd is extensively metabolized, and only a minor fraction of the radioactivity is found in DNA (6) . The latter findings have driven a search for tracers that are metabolized less readily, such as [18F]FLT (3'-deoxy-3'fluorothymidine; Ref. 7 ) and [11C]FMAU (2'-fluoro-5-methyl-1-ß-D-arabinofuranosyluracil; Ref. 8 ). It is labeled thymidine, however, that is currently the most widely studied proliferation marker for PET.
Initial attempts to use thymidine labeled in the methyl position yielded poor results because of a large number of labeled metabolites, and the method requires individual HPLC analysis of metabolites in blood samples (9) . A better approach has been the use of thymidine labeled in the 2-C position that was developed initially by Vander Borght et al. (10 , 11) and subsequently by Shields et al. (4 , 12 , 13) . In comparison with methyl-labeled thymidine, 2-C-labeled thymidine yields fewer radioactive metabolites and yields metabolites that are less likely to be trapped in tissue and mistaken for labeled thymidine incorporated into DNA (10 , 12) . Although 2-C-labeled thymidine is metabolized in vivo at the same rate as the methyl derivative, [11C]CO2 is the main metabolite (11) . Much of the labeled CO2 is rapidly eliminated from the lungs (14) , and what remains can be analyzed using developed methodology (15, 16, 17) . In addition, kinetic modeling techniques have been developed that attempt to account for the confounding influence of the tissue metabolites (4 , 18 , 19) . Preliminary studies have already described the potential of the method to measure proliferation (10) and tumor response to therapy (4) .
Despite the publication of studies demonstrating the potential of 2-[11C]thymidine PET, there are no reports correlating 2-[11C]thymidine PET-derived parameters with conventional measures of proliferation in human tumors. The aim of the study reported here, therefore, was to further explore 2-[11C]thymidine PET methodology as a marker of human tumor proliferation. The study was designed to measure tumor proliferation using 2-[11C]thymidine in a series of patients with advanced intra-abdominal malignancies and correlate PET-derived parameters with a well-established immunohistochemical measure of tumor proliferation, the Ki-67 index (20) .
|
MATERIALS AND METHODS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Production of Radiotracers.
Production of [15O]CO2 was carried out by deutron bombardment of a 14N target in a Scanditronic MC40 cyclotron. The gas was then exposed to a mixture of N2 and CO2 to produce [15O]CO2. All gaseous CO2 produced was analyzed for radiochemical purity and sterility before administration. A detailed description of the method for producing 2-[11C]thymidine has been given elsewhere (21)
. Formulations for i.v. injection were analyzed by HPLC to ensure chemical and radiochemical purity and by the Limulus amoebocyte lysate test to ensure apyrogenicity. Random samples were shown to be sterile.
Patient Imaging.
Imaging was performed on an ECAT 931-08/12 scanner (CTI/Siemens, Knoxville, TN) after insertion of arterial and i.v. lines for blood sampling and injection of radiotracer, respectively. Patients were positioned on the scanner using skin markings of bony landmarks that were obtained using X-ray simulation of a ROI including the tumor that was defined from a CT scan. CT-PET image coregistration was not possible because of poor image alignment in the abdomen. To derive quantitative image data, results were calibrated and corrected for tissue attenuation and dead time. A 68Ge phantom was used to calibrate the PET data with radioactivity measured in a well counter and enabled the image data to be expressed in kBq·ml-1. An initial 20-min transmission scan was obtained for correction of tissue attenuation. In a subset of patients, the transmission scan was followed by a 10-min perfusion scan using inhaled [15O]CO2 (activity = 4 MBq) and a constant flow rate of 500 ml·min-1 for 210 s, starting 30 s into the scan. Arterial blood was withdrawn at a rate of 5 ml·min-1, and radioactivity due to 15O was monitored continually by passing the arterial whole blood through an on-line bismuth germinate detector cross-calibrated with a well counter (16
, 22)
.
The 2-[11C]thymidine scan was performed approximately 15 min after the perfusion scan. 2-[11C]Thymidine was administered i.v. as a bolus over 30 s, 30 s after starting scanning. An on-line measuring system was used to record the radioactivity of arterial blood samples at 1-s intervals. Discrete blood samples were also taken up to 60 min after tracer administration to calibrate the radioactive counts, define the plasma to blood partitioning (n = 10), and measure the relative amounts of parent and metabolite compounds in the plasma (n = 6) using HPLC (23) .
PET Image Analysis.
The tomographic data were reconstructed using the method of filter back-projection. Voxel dimensions were 2.1 x 2.1 x 6.4 mm, and the radial, tangential, and axial full half width maximum values were 8.4, 8.3, and 6.6 mm, respectively. Inspection of a recent CT scan was used to assist the delineation of tumor and normal tissue regions. ROIs were defined using Analyze image analysis software (Biomedical Imaging Resource, Mayo Foundation) on integral images. To ensure only viable tumor was analyzed, tumor rims were defined using the conventional images, the perfusion scans, and the 2-[11C]thymidine data. To minimize the effect of organ movement, tumor volume was defined on a minimum of five planes, avoiding those at the top and bottom and those at the apparent edge of an organ. All tumors greater than 4 cm2 contained areas of central necrosis that were excluded from the ROIs (patients 4–7, 9–11, and 14–17). Once the ROIs were defined, they were applied to the dynamic images to generate time-activity curves for the tumor regions.
Quantification of Tissue Tracer Uptake.
Changes in the imaged tumor tracer concentration measured over time, together with the plasma data, were used to estimate physiological parameters. In a subset of patients given [15O]CO2, values for tissue perfusion (ml·min·mg-1) were obtained using the model described originally by Kety and Schmidt (24)
. A SUV was calculated for each time frame [SUV = measured tissue activity/(injected activity/patient weight)] from the start (t = 0) to the end (t = 3600 s) of the scan. The data were analyzed to give SUV3000–3600 (kg·ml-1) calculated as the midpoint between the SUV obtained at 3000 and 3600 s. The integral of the SUV time curve was used to calculate the area under the time-activity curve, AUC3600 (kg·min·ml-1).
Kinetic modeling of the PET activity versus time data was carried out using spectral analysis, which fits a large number of discrete exponential curves to the data with no predefined compartments. Spectral analysis was used to calculate the tumor IRF using a plasma metabolite corrected input function and the corresponding tumor time-activity curve (15) . Values for IRF were obtained for the delivery (IRF1 min) and the uptake (IRF60 min) of tracer in tumor. The FRT in tissue at 1 h was calculated as IRF60 min/IRF1 min. FRT values can range from 0–1, corresponding with 0–100% of the delivered tracer being retained in a tissue at 60 min.
Ki-67 Immunostaining.
Before PET scanning, samples of the tumor to be imaged (the primary disease or a synchronously occurring metastasis) were taken for immunocytochemical analysis by MIB1 antibody staining. The antibody staining was performed on paraffin-embedded sections of tumor using published methodology (20)
. Briefly, after dewaxing and blocking endogenous peroxidase with 3% H2O2 for 20 min, sections were rinsed and immersed in boiling 10 mM citric acid (pH 6) and placed in a pressure cooker at 15 p.s.i. for 1 min. After rinsing further, sections were incubated with normal rabbit serum diluted 1:5 in PBS for 5 min, followed by 1 h with MIB1 antibody (Immunotech, S.A., Marseilles, France). Visualization of the antibody was obtained by incubation with conjugated streptavidin/biotin complex for 30 min followed by 0.5% diaminobenzidine and 0.03% H2O2 in PBS for 2–7 min. Finally, sections were counterstained with hematoxylin. Cell counting was performed without prior knowledge of the PET scan results. For each section, 2000 tumor cells were counted, and the number of nuclear-labeled tumor cells was expressed as a percentage of the total to provide the Ki-67 labeling index. The relationships between variables were investigated using the Pearson correlation coefficient.
|
RESULTS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
and 2
list the patient characteristics and PET variables. The mean age and weight were 65 years and 65 kg, respectively. The mean values with SDs for the injected radiotracer dose, radiochemical purity, and specific activity were 336 ± 145 MBq (range, 80–547 MBq), 99 ± 0.8% (range, 97.3–99.7%), and 19378 ± 13786 MBq/µmol (range, 5433–63060 MBq/µmol), respectively. The volume of tumor tissue imaged ranged from 2.8 to 17.6 cm3 (mean, 7.5 cm3).
|
|
illustrates a typical time course for the distribution of 2-[11C]thymidine in plasma. After a bolus injection of 2-[11C]thymidine, the plasma radioactivity increased rapidly and reached a peak approximately 60 s after tracer administration. This was followed by a rapid fall in activity to approximately 5% of the starting value 5 min after injection. The percentage concentration of [11C]CO2, the predominant labeled metabolite, reached a plateau by 900 s and contributed 65–70% of all of the 11C detected.
|
. Tracer distribution was pronounced and most homogeneous in the tumor rims, aiding the definition of ROIs. ROIs drawn in normal tissues showed a rapid delivery of the tracer in the liver and spleen followed by slow washout, with faster clearance in the spleen. Rapid accumulation of 2-[11C]thymidine was seen in the kidney followed by minimal clearance.
Analysis of the Influence of Tissue Perfusion on PET Data.
Table 2
lists the tumor PET data obtained from the scans. Values for the FRT ranged from 0.096 to 0.690, with a mean of 0.343. In a subset of 11 patients, tumor blood flow ranged from 0.17 to 0.97 ml·min·mg-1, with a mean of 0.52 ml·min·mg-1. There was no significant relationship between tumor blood flow and the FRT (r = -0.40; Fig. 2
).
|
, with a mean of 42%. Fig. 3
illustrates the relationship between Ki-67 index and 2-[11C]thymidine PET measurements of tracer incorporation. No significant relationship was seen between the Ki-67 index and either SUV or AUC. The IRF60 min was also studied as a measure of the retention of the tracer. Although there was a correlation between IRF60 min and Ki-67 index (r = 0.68; P = 0.01), it was skewed by two outlying points. There was a significant correlation between Ki-67 index and FRT (r = 0.58; P = 0.02) that was not influenced by outlying data points.
|
. The median interval between biopsy and PET scan was 4 weeks (range, 0.1–126 weeks; Table 1
). Biopsies for Ki-67 staining were obtained from either nonimaged (n = 7) or imaged (n = 10) tumor. The correlation between the in vivo PET and ex vivo immunohistochemical measurement of tumor proliferation was not influenced by the timing of the biopsy (<4 weeks or >4 weeks before PET scan), the origin of the biopsy for Ki-67 staining (primary or metastasis), or whether the biopsy was from imaged or nonimaged tumor.
|
|
DISCUSSION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The analysis demonstrated no significant correlation between Ki-67 index and either SUV or AUC. The AUC is strongly influenced by data collected at an early time point because of the fall in activity within the ROI. This makes the AUC dependent on tracer delivery. Similarly, SUV does not correct for differences in the delivery of the tracer. The full potential of the PET technique requires the use of an appropriate model to relate the observed time course of the tracer in tissue to the concentration of tracer in plasma. The conventional approach to tracer kinetic modeling of PET data is to use a compartmental model with different compartments representing a particular tissue space or chemical form of the compound. Compartmental models require some knowledge of the biological fate of the compounds in vivo, and, because the number of compartments and relationships between them are fixed, they are highly constrained. If the assumptions are incorrect, then the model will be inaccurate. The model described by Mankoff et al. (18) included five tissue compartments requiring the estimation of eight rate constants from three measured blood input functions (thymidine, intermediate metabolites, and CO2) and the time-activity curve. The latter approach can lead to a problem in identifying parameters. It was anticipated that it would not be possible to estimate all parameters accurately due to the similarity in time courses for each input function and the inherent noise in the tissue data. Spectral analysis (26) was therefore used to obtain pharmacokinetic parameters because it requires no a priori decision about the most appropriate compartmental structure (26, 27, 28) . Spectral analysis allows for the characterization of the IRF of the tracer and accounts for the interpatient variability in tracer delivery.
The FRT parameter is related to the mean transit time of the tracer that equals the volume of distribution/delivery. The FRT is more robust to partial volume effects than other parameter estimates. Partial volume effects result from the limited spatial resolution of the tomograph. Thus, if a region contains tissue that is not homogeneous (for instance, where much of it is necrotic and lacks a signal), then the SUV and AUC parameters will underestimate the level of tracer incorporation. The FRT, however, will be independent of these partial volume effects. The FRT, like the other parameters, will still be susceptible to partial volume effects that relate to the spill in and out of regional signals due to the tomograph’s inherent point spread function.
A relationship was seen between the level of tracer retained in tumors, measured using the spectral analysis, and Ki-67 index. Although the spectral analysis offers an improvement over SUV and AUC, it is important to note that none of the parameters obtained attempted to correct for the contribution of labeled metabolites in tissue. [11C]CO2 has been shown to comprise around 15% of the tumor activity (15) , and the contribution of the other labeled metabolites is unknown. A method, based on a dual-scan approach, was developed to correct for the presence of radiolabeled metabolites (15) , but only after the data presented in this study were collected. It is likely, therefore, that there may be potential for improving the level of correlation between Ki-67 index and the in vivo measure of proliferation if further measurements of the metabolite contribution to the tissue could be made.
The IRF60 min/IRF1 min ratio (FRT) appeared more robust than using IRF60 min. The correlation between Ki-67 index and FRT was similar to those reported for Ki-67 index and tritiated thymidine (29 , 30) , BrdUrd (31, 32, 33) , and IUdR (34) indices. The wide range of Ki-67 indices seen in the cohort of patients studied was representative of a heterogeneous patient group, reflected by a similar range in values for FRT. The difference in tissue sampling techniques between the PET and pathology methods will contribute to the modest correlation between the parameters. For example, a tumor with a small number of highly proliferative cancer cells and a large number of nonproliferating normal cells (e.g., fibroblasts, endothelial cells, macrophages, and lymphocytes) might appear less active using PET than a tumor with a high proportion of more slowly proliferating tumor cells and a small number of normal cells. A number of physiological and biological properties of tumors may also negatively influence the relationship between FRT and Ki-67 index. Some of the tissue samples were obtained many weeks before the PET study at the time of initial diagnosis. The biological variability between primary tumors and metastases may also be a confounding factor because of heterogeneity in the number of proliferating and metastasizing cells (35) . In addition to the biological heterogeneity of these tumors that could potentially mask any true differences in proliferative behavior, physiological factors such as tumor perfusion may be important. The level of correlation seen between the Ki-67 index and FRT in the group of patients studied, despite the above-mentioned potentially confounding factors, supports the continued development of the methodology as a measure of human tumor proliferation.
In summary, this study has shown a statistically significant correlation between tumor FRT obtained in vivo using 2-[11C]thymidine and proliferative index measured ex vivo using Ki-67 staining. The finding supports the use of 2-[11C]thymidine PET as a means of measuring changes in tumor proliferation in vivo in response to antiproliferative treatment. This may be potentially useful in the selection or modification of treatment based on the individual biological characteristics of a tumor with the aim of improving treatment outcome.
|
FOOTNOTES |
|---|
1 Supported by grants from the Medical Research Council of the United Kingdom and Cancer Research UK. 
2 To whom requests for reprints should be addressed. Present address: Molecular Imaging Centre, Academic Department of Radiation Oncology, Christie NHS Trust Hospital, Wilmslow Road, Manchester M20 4BX, United Kingdom. Phone: 044-0-161-446-8003; Fax: 044-0-161-446-8111; E-mail: Anne.Mason{at}christie-tr.nwest.nhs.uk
3 The abbreviations used are: PET, positron emission tomography; FDG, fluorodeoxyglucose; IUdR, iododeoxyuridine; HPLC, high-pressure liquid chromatography; ROI, region of interest; SUV, standard uptake value; AUC, area under the curve; IRF, impulse response function; FRT, fractional retention of thymidine; BrdUrd, bromodeoxyuridine; CT, computed tomography.
Received 4/28/02. Accepted 8/15/02.
|
REFERENCES |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  R. Ullrich, H. Backes, H. Li, L. Kracht, H. Miletic, K. Kesper, B. Neumaier, W.-D. Heiss, K. Wienhard, and A. H. Jacobs Glioma Proliferation as Assessed by 3'-Fluoro-3'-Deoxy-L-Thymidine Positron Emission Tomography in Patients with Newly Diagnosed High-Grade Glioma Clin. Cancer Res., April 1, 2008; 14(7): 2049 - 2055. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. K. Buck, M. Bommer, S. Stilgenbauer, M. Juweid, G. Glatting, H. Schirrmeister, T. Mattfeldt, D. Tepsic, D. Bunjes, F. M. Mottaghy, et al. Molecular Imaging of Proliferation in Malignant Lymphoma. Cancer Res., November 15, 2006; 66(22): 11055 - 11061. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Perumal, R. G. Pillai, H. Barthel, J. Leyton, J. R. Latigo, M. Forster, F. Mitchell, A. L. Jackman, and E. O. Aboagye Redistribution of Nucleoside Transporters to the Cell Membrane Provides a Novel Approach for Imaging Thymidylate Synthase Inhibition by Positron Emission Tomography. Cancer Res., September 1, 2006; 66(17): 8558 - 8564. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N. Gupta, A. Saleem, B. Kotz, S. Osman, E. O. Aboagye, R. Phillips, C. Vernon, H. Wasan, T. Jones, P. J. Hoskin, et al. Carbogen and Nicotinamide Increase Blood Flow and 5-Fluorouracil Delivery but not 5-Fluorouracil Retention in Colorectal Cancer Metastases in Patients. Clin. Cancer Res., May 15, 2006; 12(10): 3115 - 3123. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A Saleem Potential of PET in oncology and radiotherapy Br. J. Radiol., November 1, 2005; Supplement_28(1): 6 - 16. [Full Text] [PDF]
|
|
|
|
|
|
P. Wells, R. N. Gunn, C. Steel, A. S. Ranicar, F. Brady, S. Osman, T. Jones, and P. Price 2-[11C]Thymidine Positron Emission Tomography Reproducibility in Humans Clin. Cancer Res., June 15, 2005; 11(12): 4341 - 4347. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. F. Shields PET Imaging with 18F-FLT and Thymidine Analogs: Promise and Pitfalls J. Nucl. Med., September 1, 2003; 44(9): 1432 - 1434. [Full Text] [PDF]
|
|
|
|
|
|
H. Barthel, M. C. Cleij, D. R. Collingridge, O. C. Hutchinson, S. Osman, Q. He, S. K. Luthra, F. Brady, P. M. Price, and E. O. Aboagye 3'-Deoxy-3'-[18F]Fluorothymidine as a New Marker for Monitoring Tumor Response to Antiproliferative Therapy in Vivo with Positron Emission Tomography Cancer Res., July 1, 2003; 63(13): 3791 - 3798. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. Wells, E. Aboagye, R. N. Gunn, S. Osman, A. V. Boddy, G. A. Taylor, I. Rafi, A. N. Hughes, A. H. Calvert, P. M. Price, et al. 2-[11C]Thymidine Positron Emission Tomography as an Indicator of Thymidylate Synthase Inhibition in Patients Treated With AG337 J Natl Cancer Inst, May 7, 2003; 95(9): 675 - 682. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Cancer Research | Clinical Cancer Research |
| Cancer Epidemiology Biomarkers & Prevention | Molecular Cancer Therapeutics |
| Molecular Cancer Research | Cancer Prevention Research |
| Cancer Prevention Journals Portal | Cancer Reviews Online |
| Annual Meeting Education Book | Meeting Abstracts Online |
