HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) All Versions of this Article: 2393041382v1 239/3/740    most recent Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Stewart, E. E. Articles by Lee, T.-Y. Search for Related Content PubMed PubMed Citation Articles by Stewart, E. E. Articles by Lee, T.-Y.

Correlation between Hepatic Tumor Blood Flow and Glucose Utilization in a Rabbit Liver Tumor Model¹

¹ From the Lawson Health Research Institute, St Joseph's Health Care London, London, Ontario, Canada (E.E.S., J.H., T.Y.L.); Department of Medical Biophysics, Faculty of Medicine and Dentistry, University of Western Ontario, London, Ontario, Canada (E.E.S., T.Y.L.); and Imaging Research Laboratories, Robarts Research Institute, 100 Perth Dr, London, ON, Canada N6A 5K8 (E.E.S., X.C., J.H., T.Y.L.). Received August 10, 2004; revision requested October 19; final revision received July 6, 2005; final version accepted August 1. Supported by the Canadian Institutes of Health Research. E.E.S. supported in part by GE Healthcare. Address correspondence to T.Y.L. (e-mail: tlee{at}imaging.robarts.ca).

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX References

 
Purpose: To prospectively determine the relationship between hepatic tumor blood flow and glucose utilization in vivo by using a combined positron emission tomographic (PET)/computed tomographic (CT) scanner.

Materials and Methods: The animal care and use subcommittee at the University of Western Ontario approved this study. VX2 carcinoma cells were implanted in the livers of eight male New Zealand white rabbits. Functional CT was performed before tumor implantation and every 4 days thereafter. Each examination consisted of two phases: In the first phase, 30-second cine breath-hold scanning was performed with simultaneous injection of 5 mL of contrast material. In the second phase, 4-second cine scanning was performed without breath holding every 10 seconds for 2 minutes. Second-phase CT images were coregistered with first-phase images to eliminate breathing artifacts. The weighted summation of the aortic and portal venous time-attenuation curves was deconvolved against curves from the liver to derive hepatic blood flow (HBF). Five animals underwent fluorine 18 fluorodeoxyglucose (FDG) scanning before and every 8 days after implantation. FDG uptake was measured as standardized uptake value (SUV). Data were analyzed with repeated-measures analysis of variance and the Tukey-Kramer multiple comparison test. Linear regression was used to compare SUV and HBF in tumors and normal tissue.

Results: In the hypovascular tumor core, (a) mean HBF decreased from 262 mL · min^–1 · 100 g^–1 ± 22 (standard deviation) at baseline to 101 mL · min^–1 · 100 g^–1 ± 62 at the end of the study (P < .05) and (b) mean SUV increased from 2.12 g/mL ± 0.06 to 4.56 g/mL ± 0.73 (P < .05) during the same period.

Conclusion: Functional CT in combination with FDG PET can be used to observe changes in HBF and glucose utilization in a growing liver tumor.

© RSNA, 2006

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX References

 
As the metabolic demand of a tissue increases, an equivalent increase in oxygen supply by means of increased blood flow (perfusion) is necessary to maintain aerobic respiration (1,2). The coupling of blood flow to metabolic rate under physiologic and pathophysiologic conditions has been demonstrated in a variety of organs (the brain and heart) and tissues (muscle), as well as in various cancers (2–6).

Malignant neoplasms usually exhibit regions of low oxygenation due to poor perfusion (7). The limited availability of oxygen in these hypoxic regions results in increased anaerobic glycolysis and hence less efficient use of glucose, which in turn necessitates high extraction efficiency and uptake of glucose and its radioactive analogue, fluorine 18 fluorodeoxyglucose (FDG). Thus, autoradiographic studies in animal models have demonstrated that there is always increased FDG uptake adjacent to the necrotic area of a tumor (8,9). More specifically, Mankoff et al (4) suggested that an elevated ratio of FDG uptake to blood flow is an indicator of glucose utilization relative to delivery and that an elevated ratio or uncoupling of blood flow and glucose utilization indicates high glucose extraction by the tumor. Besides indicating higher glucose extraction by the tumor, the uncoupling of blood flow and glucose utilization may also indicate tissue hypoxia. Clinical data have shown that hypoxia negatively affects the treatment outcome of both radiation therapy and anticancer drug treatment (10). There is substantial potential for therapeutic benefit if tumor hypoxia can be diagnosed with noninvasive methods and the information can be used to identify high-risk populations and modify therapy accordingly.

Functional computed tomography (CT) (11) and FDG positron emission tomography (PET) are two functional imaging methods that can be used to noninvasively measure physiologic changes (specifically, blood flow and glucose utilization in the liver). The measurement of FDG metabolic rate in the liver, however, is relatively invasive (12). Hence, the accumulation of FDG—which can be measured with the standardized uptake value (SUV)—is considered to reflect the rate of tumor glycolytic metabolism (13). Changes in liver perfusion correlate with the growth of new tumor blood vessels (angiogenesis) (14). Researchers have shown that dynamic contrast material–enhanced CT is useful in the noninvasive measurement of liver perfusion (15,16).

In contrast to other organs or tissues, which are supplied only by arteries, the liver has a dual blood supply. The liver receives its blood from both the hepatic artery, which delivers oxygenated blood from the heart, and the portal vein, which drains venous blood from the gastrointestinal tract (17). Approximately two-thirds of the blood flow to the liver is supplied by the portal vein; the remaining one-third is supplied by the common and proper hepatic arteries (18). The purpose of our study was to prospectively determine the relationship between hepatic tumor blood flow and glucose utilization in vivo by using a combined PET/CT scanner.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX References

 
Study Protocol
Eight healthy male New Zealand white rabbits (weight, 2.9–3.5 kg) were used in these experiments. The animal care and use subcommittee at the University of Western Ontario approved the experimental procedure. Partial financial support was provided by GE Healthcare (in the form of salary support for E.E.S.); however, the authors controlled the data and information submitted for publication. In each animal, anesthesia was induced with isoflurane administered through a mask, an endotracheal tube was inserted, and anesthesia was maintained by means of mechanical ventilation at a rate of 20 breaths per minute with a mixture of oxygen and isoflurane (1.5%–2.5%). Respiratory reflex was suppressed with vecuronium bromide (if required). The animal's body temperature was maintained between 38.5°C and 39.5°C with a heated recirculating water blanket.

Tumor cells were harvested from a VX2 carcinoma that was grown in a leg of a donor rabbit. Subsequently, these harvested cells (0.25 mL) were suspended in Hanks balanced salt solution (Sigma-Aldrich Canada, Oakville, Ontario, Canada) and injected directly into the liver of the rabbit (J.H., E.E.S.). Functional CT was performed before tumor implantation and every 4 days thereafter until metastatic nodules were observed in the lung (on average, 24 days after injection). At that time, the animals were sacrificed with an overdose of sodium pentobarbital. Five of the animals underwent FDG PET before tumor implantation and every 8 days thereafter.

Functional CT Protocol
One ear vein was cannulated for administration of vecuronium bromide, iohexol (GE Healthcare, Waukesha, Wis), and FDG (Hamilton Health Science Center, Hamilton, Ontario, Canada) with use of the synthesis technique described by Hamacher and Coenen (19). The abdomen of the rabbit was scanned with the CT component of the combined PET/CT scanner (Discovery LS; GE Healthcare, Waukesha, Wis) by two authors (E.E.S., J.H.) with the following technique: Localization transverse CT was performed to position four 5-mm sections. The first section was positioned at the dome of the liver just below the diaphragm, whereas the fourth section was positioned at the level of the upper pole of either kidney.

To determine liver perfusion, CT images of the four chosen sections were acquired with a two-phase scanning protocol. The scanning parameters for the first phase were 80 kVp, 60 mA, and 1 second per rotation, with four 5-mm sections acquired per rotation. During this phase, the liver was continuously (cine) scanned for 30 seconds. A nonionic contrast material bolus (concentration, 200 mg iodine per milliliter; dose, 1.5 mL per kilogram of body weight) was simultaneously injected at a rate of 1 mL/sec. Before contrast material injection, the ventilator was turned off to eliminate motion artifacts due to breathing. The images were reconstructed at 0.5-second intervals to provide time-attenuation curves of the same time interval.

The second phase of the scanning protocol started 10 seconds after the first phase was completed. The mechanical ventilator was turned on and used to maintain respiration at a rate of 20 breaths per minute, and a 4.0-second burst of cine scanning was performed every 10 seconds for a period of 2 minutes. The scanning parameters were 80 kVp, 60 mA, and 1 second per rotation, with four 5-mm sections acquired per rotation. In this phase, however, images were reconstructed at 0.2-second intervals instead of 0.5-second intervals; this resulted in 19 images per section location for each cine scan.

FDG PET Protocol
The FDG PET images were acquired (E.E.S., J.H.) after functional CT. Prior to functional CT, 50 MBq ± 15 (standard deviation) of FDG was administered by means of an ear vein catheter. Approximately 40 minutes later, four-section helical CT covering the thorax and abdomen was performed with a 1-second rotational speed, 1.5:1.0 pitch, and 50-cm transaxial field of view. The x-ray tube was operated at a voltage of 140 kVp and a current of 80 mA. CT images were reconstructed into 5-mm-thick sections at 4.25-mm intervals and were subsequently used to correct the PET emission images for attenuation. At least three of the CT images obtained with this method matched three of the 5 x 5-mm sections obtained with functional CT. The PET scanner had a 55-cm transaxial field of view and a 15.2-cm transverse field of view divided into 35 4.25-mm-thick sections, which coincided with the helical CT images. PET images were acquired in the two-dimensional mode (septa between detector rings of the scanner were fully extended to reduce cross-plane scatter) for 5 minutes per transverse field of view for a total of 10 minutes (ie, two bed positions were needed to scan the entire thorax and abdomen of a rabbit).

Software provided with the PET/CT scanner was used to initially correct the PET emission data for random coincidences, dead time, scatter coincidences, and attenuation by using CT attenuation correction maps obtained with helical CT. The corrected data were reconstructed into transaxial images by using ordered-subset expectation maximization software. The resulting in-plane image resolution of the transaxial images was approximately 4.5 mm full width at half maximum, with a transverse resolution of approximately 4.5 mm full width at half maximum.

Functional Parameters Calculation
Images obtained with first- and second-phase functional CT scans were coregistered (E.E.S.) to remove motion artifacts due to breathing. From each 4.0-second burst of cine scanning during the second phase, the image that best matched the images obtained at the same section location in the first phase of the examination (images obtained with breath holding) was selected. The selected second-phase images were used together with the first-phase breath-hold images to calculate the functional parameters (Appendix).

Contrast enhancement or time-attenuation curves of the aorta (to approximate the hepatic artery), portal vein, and liver parenchyma were obtained from the coregistered first- and second-phase images. A weighted summation of the aortic and portal venous curves was deconvolved against the liver parenchymal curve by using the model described by Johnson and Wilson (20) with a software program developed in our laboratory (T.Y.L., X.C.). The sum of squared deviations of the fitted curve from the measured liver parenchyma or tumor contrast enhancement curves was minimized to determine the optimal weights of the aortic and portal venous curves, as well as the optimal impulse residue function according to the model described by Johnson and Wilson (20) (Appendix). The determined impulse residue function is used to calculate the following functional parameters: hepatic blood flow (HBF), hepatic blood volume (HBV), capillary permeability surface area product (PS), appearance time of contrast material in tissue relative to that in the aorta (T₀), and hepatic arterial fraction (HAF) or the fraction of HBF that is derived from the hepatic artery as opposed to the portal vein (Appendix). Functional maps were generated with deconvolution, as described, above the liver or tumor enhancement curve corresponding to each voxel.

Data Analysis
From the set of five functional maps, four regions of interest (ROIs) were drawn (E.E.S., T.Y.L.) to measure perfusion and other available functional parameters (HBV, PS, T₀, and HAF) in the tumor and adjacent normal tissue. The first ROI was drawn around normal tissue on HBF maps. The second ROI was drawn around the tumor core on areas with an HBF of less than or equal to 75% of that measured in the ROI around normal tissue. The corresponding T₀ map was used to draw a third ROI covering the entire tumor, which comprised the tumor core and adjacent tumor tissue. This ROI was drawn to include voxels where T₀ was less than or equal to half of that in the normal voxel. A fourth ROI was drawn around the tumor rim and was defined as the difference between the ROI of the tumor core and the ROI of T₀. Measurements were obtained by using ROIs drawn around the tumor core, tumor rim, and normal tissue. The reported values reflect averages from between two and four functional CT sections that cover a 1–2-cm slab through the tumor and normal liver parenchyma.

For quantitative analysis of the acquired FDG PET images, ROIs were drawn around areas with enhanced FDG uptake relative to the adjacent normal tissue, which also encompassed the entire tumor in the corresponding CT image. These regions were representative of the most metabolically active portion of the tumors; thus, they likely represented areas with the most biologically aggressive behavior. FDG uptake was measured in all ROIs as whole-body SUV (unit of measure, grams per milliliter), and the mean SUV over all sections covering the tumor was reported.

Statistical Analysis
Statistical analysis was performed by using SAS software (SAS Proc Mixed for Windows, release 8.2; SAS Institute, Cary, NC); to account for missing data, repeated-measures analysis of variance was used. We used a completely within-subjects design, where time after initial detection and tissue type were the independent variables and CT functional parameters or SUV were the dependent variables. Tukey-Kramer multiple-comparison tests were used for post hoc comparisons, with a P value of less than .05 considered to indicate a significant difference. Linear regression analysis was used to compare the SUV measured on FDG PET images to the HBF values derived from functional CT. To investigate the sensitivity of our results relative to the choice of thresholds, a series of different thresholds were investigated (E.E.S.). These thresholds were as follows: (a) 75% of normal HBF, 50% of normal T₀; (b) 65% of normal HBF, 40% of normal T₀; and (c) 85% of normal HBF, 60% of normal T₀. These data were also analyzed with the statistical software: Time after initial detection, thresholds, and tissue type were the independent factors, and functional CT parameters were the dependent variables.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX References

 
Contrast enhancement curves generated from the aorta, portal vein, normal tissue, and tumor show that contrast enhancement appears much earlier in the tumor than in the adjacent normal tissue (Fig 1). The time taken for contrast material to arrive at the tumor relative to T₀ decreased by a mean of 64.9% ± 7.8 in the tumors when compared with that in the adjacent normal liver tissue (data not shown). We used the T₀ map and contrast-enhanced CT images to detect tumors as early as 4 days after they were implanted. However, since the time of initial detection varied among the rabbits (between 4 and 12 days after implantation), data were recorded at baseline and each day after initial detection of the tumor. We were able to detect relatively small tumors with a mean average diameter of 0.76 cm ± 0.14 at the time of initial detection; these tumors had grown to a mean diameter of 2.99 cm ± 0.90 by the end of the study. A comparison between FDG PET images and HBF and T₀ maps at baseline and 8 and 16 days after the initial detection of the tumor shows how these functional parameters change with tumor growth (Fig 2). On these maps, we were able to identify major blood vessels and the tumor, as well as observe the development of the hypovascular core.

View larger version (30K): [in this window] [in a new window] [Download PPT slide]   Figure 1a: Graphs show contrast enhancement curves of the aorta, portal vein, normal liver parenchyma, tumor rim, and tumor core. Measurements were obtained with CT. (a) Data obtained with first-phase breath-hold CT scanning. (b) Data obtained with coregistered first- and second-phase CT scanning. These curves were deconvolved to calculate HAF, HBF, HBV, mean transit time, PS, and T0. (c, d) Enlarged views of the curves measured in the tumor and normal liver tissue; the curve of the tumor core was multiplied by a factor of three. During the first phase of scanning (c), neither tumor enhancement nor normal tissue enhancement returned to baseline levels after peak enhancement because contrast material leaked into the extravascular space. Contrast material appeared in the tumor and aorta at the same time, whereas contrast material appeared in normal tissue and the portal vein at about the same time. This suggests the tumor derived more of its blood supply from the hepatic artery, whereas normal tissue derived more of its blood supply from the portal vein.

View larger version (28K): [in this window] [in a new window] [Download PPT slide]   Figure 1b: Graphs show contrast enhancement curves of the aorta, portal vein, normal liver parenchyma, tumor rim, and tumor core. Measurements were obtained with CT. (a) Data obtained with first-phase breath-hold CT scanning. (b) Data obtained with coregistered first- and second-phase CT scanning. These curves were deconvolved to calculate HAF, HBF, HBV, mean transit time, PS, and T0. (c, d) Enlarged views of the curves measured in the tumor and normal liver tissue; the curve of the tumor core was multiplied by a factor of three. During the first phase of scanning (c), neither tumor enhancement nor normal tissue enhancement returned to baseline levels after peak enhancement because contrast material leaked into the extravascular space. Contrast material appeared in the tumor and aorta at the same time, whereas contrast material appeared in normal tissue and the portal vein at about the same time. This suggests the tumor derived more of its blood supply from the hepatic artery, whereas normal tissue derived more of its blood supply from the portal vein.

View larger version (25K): [in this window] [in a new window] [Download PPT slide]   Figure 1c: Graphs show contrast enhancement curves of the aorta, portal vein, normal liver parenchyma, tumor rim, and tumor core. Measurements were obtained with CT. (a) Data obtained with first-phase breath-hold CT scanning. (b) Data obtained with coregistered first- and second-phase CT scanning. These curves were deconvolved to calculate HAF, HBF, HBV, mean transit time, PS, and T0. (c, d) Enlarged views of the curves measured in the tumor and normal liver tissue; the curve of the tumor core was multiplied by a factor of three. During the first phase of scanning (c), neither tumor enhancement nor normal tissue enhancement returned to baseline levels after peak enhancement because contrast material leaked into the extravascular space. Contrast material appeared in the tumor and aorta at the same time, whereas contrast material appeared in normal tissue and the portal vein at about the same time. This suggests the tumor derived more of its blood supply from the hepatic artery, whereas normal tissue derived more of its blood supply from the portal vein.

View larger version (28K): [in this window] [in a new window] [Download PPT slide]   Figure 1d: Graphs show contrast enhancement curves of the aorta, portal vein, normal liver parenchyma, tumor rim, and tumor core. Measurements were obtained with CT. (a) Data obtained with first-phase breath-hold CT scanning. (b) Data obtained with coregistered first- and second-phase CT scanning. These curves were deconvolved to calculate HAF, HBF, HBV, mean transit time, PS, and T0. (c, d) Enlarged views of the curves measured in the tumor and normal liver tissue; the curve of the tumor core was multiplied by a factor of three. During the first phase of scanning (c), neither tumor enhancement nor normal tissue enhancement returned to baseline levels after peak enhancement because contrast material leaked into the extravascular space. Contrast material appeared in the tumor and aorta at the same time, whereas contrast material appeared in normal tissue and the portal vein at about the same time. This suggests the tumor derived more of its blood supply from the hepatic artery, whereas normal tissue derived more of its blood supply from the portal vein.

View larger version (91K): [in this window] [in a new window] [Download PPT slide]   Figure 2: FDG PET images and HBF and T0 maps obtained in the transverse plane before tumor implantation and 8 and 16 days after initial detection of the tumor. Note ROIs on the normal tissue (white arrows) and tumors (red arrows) as they appear on the T0 maps and superimposed on the HBF maps. Images in the top row are combined FDG PET/CT images of a rabbit liver. Middle row images are HBF maps of the liver obtained at the same location. Sixteen days after the initial detection of the tumor, there was an HBF–glucose utilization mismatch at the tumor center, with comparatively less blood flow in the tumor center than at the tumor periphery; however, more or less uniform glycolytic activity was maintained throughout the tumor. Bottom row images are T0 maps. These images show that contrast material arrived much earlier in the tumor than in the adjacent normal tissue.

View larger version (21K): [in this window] [in a new window] [Download PPT slide]   Figure 3a: Graphs show mean (a) HBF and (b) HBV in the tumor compared with mean HBF and mean HBV, respectively, in the adjacent normal tissue. Each tumor was separated into two regions: the hypovascular tumor core and the tumor rim. For clarity, other significant differences in HBF or HBV are not shown. * = significant difference from baseline value, as calculated with two-way analysis of variance and post hoc Tukey-Kramer multiple comparison test.

View larger version (21K): [in this window] [in a new window] [Download PPT slide]   Figure 3b: Graphs show mean (a) HBF and (b) HBV in the tumor compared with mean HBF and mean HBV, respectively, in the adjacent normal tissue. Each tumor was separated into two regions: the hypovascular tumor core and the tumor rim. For clarity, other significant differences in HBF or HBV are not shown. * = significant difference from baseline value, as calculated with two-way analysis of variance and post hoc Tukey-Kramer multiple comparison test.

View larger version (23K): [in this window] [in a new window] [Download PPT slide]   Figure 4a: Graphs show (a) mean PS and (b) mean HAF as the implanted VX2 tumors grew. Each tumor was separated into two regions: the hypovascular core and the tumor rim. Normal liver tissue region values were also plotted for comparison. For clarity, other significant differences in either PS or HAF are not shown. * = significant difference from baseline value, as calculated with two-way analysis of variance and post hoc Tukey-Kramer multiple comparison test.

View larger version (23K): [in this window] [in a new window] [Download PPT slide]   Figure 4b: Graphs show (a) mean PS and (b) mean HAF as the implanted VX2 tumors grew. Each tumor was separated into two regions: the hypovascular core and the tumor rim. Normal liver tissue region values were also plotted for comparison. For clarity, other significant differences in either PS or HAF are not shown. * = significant difference from baseline value, as calculated with two-way analysis of variance and post hoc Tukey-Kramer multiple comparison test.

SUV and HBF
Both SUV (F = 15.87; df = 3, 12; P < .05; observed power = 0.999) and HBF (F = 7.94; df = 3, 12; P < .05; observed power = 0.948) in the tumor core were significantly different during the growth of the tumor. In the hypovascular tumor core, post hoc tests revealed that (a) mean SUV increased from 2.12 g/mL ± 0.06 at baseline to 4.56 g/mL ± 0.73 at the end of the study (P < .05) and (b) mean HBF decreased from 262 mL · min^–1 · 100 g^–1 ± 22 to 101 mL · min^–1 · 100 g^–1 ± 62 during the same period (P < .05) (Fig 5a). Conversely, mean SUV and HBF values in the adjacent normal liver remained relatively constant around 2.12 g/mL ± 0.27 and 291 mL · min^–1 · 100 g^–1 ± 31, respectively (P > .05) (Fig 5b). To explore the uncoupling between glucose utilization and blood flow, we plotted glucose uptake versus HBF in all the tumors from baseline to the end of the study (Fig 6). Linear regression analysis showed an inverse correlation between tumor SUV and HBF (R² = 0.727, P < .05). This shows that the spatial resolution of the PET scanner was sufficient for depicting regions of low FDG uptake in the tumor core. In Figure 7, the region of low FDG uptake was in the tumor core and had prolonged T₀ and an extremely low mean HBF of 87 mL · min^–1 · 100 g^–1 ± 4 relative to normal liver.

View larger version (23K): [in this window] [in a new window] [Download PPT slide]   Figure 5a: Graphs show mean SUV versus mean HBF in the liver as the implanted VX2 tumors grew. Error bars indicate standard deviations. (a) Comparison between HBF and SUV in tumors. As the tumors grew, HBF decreased and SUV increased; this finding indicates a mismatch between blood flow (BF) and glucose utilization. (b) Corresponding measurements in the adjacent normal tissue show that both HBF and SUV levels remained relatively constant in normal tissue over time (P > .05). * = significant difference from baseline value, as calculated with two-way analysis of variance and post hoc Tukey-Kramer multiple comparison test.

View larger version (22K): [in this window] [in a new window] [Download PPT slide]   Figure 5b: Graphs show mean SUV versus mean HBF in the liver as the implanted VX2 tumors grew. Error bars indicate standard deviations. (a) Comparison between HBF and SUV in tumors. As the tumors grew, HBF decreased and SUV increased; this finding indicates a mismatch between blood flow (BF) and glucose utilization. (b) Corresponding measurements in the adjacent normal tissue show that both HBF and SUV levels remained relatively constant in normal tissue over time (P > .05). * = significant difference from baseline value, as calculated with two-way analysis of variance and post hoc Tukey-Kramer multiple comparison test.

View larger version (18K): [in this window] [in a new window] [Download PPT slide]   Figure 6: Graph shows linear correlation between HBF and FDG uptake for all tumors from baseline to the end of the study. Note the disparity between FDG uptake and HBF, which indicates that tumors were surviving by means of anaerobic glycolysis.

View larger version (33K): [in this window] [in a new window] [Download PPT slide]   Figure 7: Transverse FDG PET scan (left) and HBF (middle) and T0 (right) maps obtained on the last day of the study. Note the areas of low glycolytic activity inside the tumor (arrows) and increased contrast material arrival time. Low FDG uptake in the tumor center suggests spatial resolution of the PET scanner was sufficient for detection of low SUV and blood flow in the tumor core.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX References

Fukumura et al (27) demonstrated that the vessel density in liver tumors was seven times lower than that in normal liver tissue and that vessel density in the tumor center was significantly lower than that in the tumor periphery. Similarly, our results show significantly higher HBV in the tumor rim than in the tumor core and lower HBV in the tumor than in normal tissue (Fig 3b). The magnitudes of the tumor HBF and HBV are likely dependent on the functional microvascular density or the perfused cross-sectional area and the velocity of flow through the vasculature. A high vascular density is a prerequisite for a high nutritive flow, but it is not necessarily indicative of this condition (28). Some tumors have diminished function because of heterogeneous distribution of blood vessels with sluggish and intermittent flow (27,28). Diminished microvascular function may lead to low HBV and may deprive the central region of the tumor of a viable blood supply; thus, it may lead to hypoxia or necrosis and the uncoupling of glucose utilization and blood flow, as we have shown in our study. This is consistent with studies that have shown a correlation between tumor size and reduced blood flow and oxygen supply (7,29).

In our study, PS in the tumor rim was found to be significantly higher than PS in the tumor core or normal tissue or PS at baseline (Fig 4). Claffey et al (30) showed a close correlation between vascular permeability factor or vascular endothelial growth factor expression, microvascular hyperpermeability, and tumor-induced angiogenesis. Conversely, low vascular permeability factor or vascular endothelial growth factor expression leads to extensive necrosis and poorly formed vasculature. Hence, the high PS found in the tumor rim in our study could have been an indication of high vascular permeability factor or vascular endothelial growth factor expression and tumor-induced angiogenesis.

Similarly, the increase in HAF seen in the tumor core was perhaps due to the recruitment of new blood vessels from the hepatic artery during angiogenesis. This indicates that the implanted VX2 carcinoma derived most of its blood supply from the arterial circulation rather than the portal circulation; this finding is in agreement with the observations of Miles et al (15). Studies in tumors have shown a linear relationship between oxygen consumption and oxygen availability (7,29). Any increase in hepatic artery flow may lead to an increased oxygen supply and therefore an increased metabolic rate. That is, there is a redistribution of HBF toward more oxygenated blood from the hepatic artery rather than toward deoxygenated blood from the portal vein. As the tumor growth accelerated, angiogenesis could not maintain an adequate supply of blood to the tumor core, consequently contributing to the uncoupling of glucose utilization and blood flow.

The locations of the four functional CT sections were directly transferable to the PET acquisitions because the combined PET/CT scanner uses a common scanning table and because we did not disturb the position of the rabbit between scans. The FDG PET examinations revealed an uncoupling of glucose utilization and tumor HBF (Fig 6). As the tumors grew, SUV increased and tumor HBF decreased, whereas SUV and HBF in normal tissue remained relatively constant (Fig 5b).

Researchers have suggested that the increase in glucose metabolism in the tumor core is due to increased glycolytic enzyme activity (31,32). Monakhov et al (33) reported that hexokinase played an important role in this process. The disparity between glucose SUV and HBF in tumors suggests that there was a limited supply of oxygen and that tumor cells in this liver tumor model were able to survive because of anaerobic glycolysis. These results are consistent with the findings of rodent studies, where blood flow rates in most isotransplanted rodent tumors decreased as tumor size increased (34–36). Recently, Fukuda et al (37) showed that a negative correlation existed between tumor glucose utilization, which was measured with SUV, and tumor blood flow, which was measured with oxygen 15–labeled water and dynamic PET scanning, in patients with hepatocellular carcinoma and those with metastatic colon cancer. This finding, however, is contrary to the findings of Mankoff et al (4) and Zasadny et al (38), who reported an increase in blood flow with an increase in tumor metabolic rate. On the other hand, Mankoff et al (4) stated that the relationship between blood flow and tumor metabolic rate showed considerable dispersion and was highly variable.

Our study had several limitations. The most important limitation was the relatively high blood glucose level (mean, 5.6 mmol/L ± 0.8 at the start of each FDG PET scan). In clinical studies, patients fasted from 4–6 hours before the examination to reduce competition from plasma glucose, thereby optimizing and standardizing tumor FDG uptake (39). It is, however, impractical for rabbits to fast to stabilize HBF and blood glucose level, as fasting causes severe peristaltic movement, which causes severe motion artifacts on CT images. The high blood glucose level may have caused poor contrast enhancement between SUV in the tumor and that in normal tissue, and this may have limited our ability to differentiate between the tumor rim and the tumor core with FDG PET (40). However, our data suggest that even without fasting, the spatial resolution of the PET scanner was sufficient to detect low SUV in the tumor core (Fig 7). This suggests that delivery of contrast material and glucose to the tumor core was impeded, possibly because of necrosis.

Partial volume averaging is an inherent source of error when imaging small objects (25). It can cause underestimation of contrast enhancement, hence leading to errors in the estimation of perfusion and related parameters in small objects. In our study, however, at the time of the initial detection of the tumors, HBF, HBV, and HAF in the tumor core were significantly different from HBF, HBV, and HAF in the adjacent normal tissue and tumor rim (Figs 3, 4b). This suggests that the spatial resolution used in this functional imaging technique enabled the tumor rim to be depicted separate from the normal tissue and the tumor core to be depicted separate from the tumor rim. Thus, the size limitation for regional functional imaging is probably between 0.59 and 1.03 cm in diameter, which was the range of sizes of the initially detected tumors in this study.

In our study, 75% of normal HBF and 50% of normal T₀ were used as thresholds to delineate the tumor core from the rest of the tumor. The mean HBF and T₀ used for the thresholds were 220 mL · min^–1 · 100 g^–1 ± 12, and 2.1 sec ± 0.1, respectively. The HBF threshold was outside of the 95% confidence interval of the HBF measured in the tumor core throughout the study period (Fig 3a). This suggests that the low HBF measured in the tumor core was not an artifact introduced by the arbitrary choice of a low threshold. Moreover, analysis of the sensitivity of the thresholds showed no significant differences (P > .05) between perfusion and related functional parameters measured with ROIs created from thresholds of 75% HBF and 50% T₀ and those measured with regions from the other thresholds.

In conclusion, there was a mismatch between tumor blood flow and glucose utilization in the tumor core. A possible interpretation of this result is that the tumor core was dependent on anaerobic glycolysis for survival; hence, it could have been hypoxic. This possibility remained unproved in our study, however, because we did not evaluate hypoxia with other hypoxic markers or direct oxygen tension measurements.

Practical Application: Our results indicate that functional CT has the potential to enhance the diagnostic ability of combined PET/CT scanners by allowing HBF to be measured without the need for a cyclotron or the on-site production of water labeled with oxygen 15, which is the radiotracer frequently used to measure blood flow (41). The added functional CT examination is inexpensive and noninvasive (except for intravenous injection of contrast material) and permits both perfusion and related functional and anatomic information pertaining to liver tumors or disease to be obtained in the same examination. Future studies should include more direct comparison between FDG PET, functional CT, and immunohistochemical hypoxic markers, such as pimonidazole hydrochloride or other bioreductive agents (42) and the vascular endothelial cell marker CD31 (43).

	APPENDIX

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX References

 
Liver perfusion maps were calculated by using the tracer kinetic model, which was first described by Johnson and Wilson (20). St Lawrence and Lee (21) derived the adiabatic approximation solution, which simplifies the calculation of perfusion maps for the liver. Since capillaries in the liver are permeable to contrast material, the model divides the liver into two principal spaces: the intravascular space and the extravascular space (20). These spaces are separated by the permeable capillary endothelium. The model uses three basic assumptions to arrive at a solution. First, the permeable capillary endothelium allows bidirectional diffusion of contrast material between the intra- and extravascular spaces. Second, there is a transverse concentration gradient of contrast material in the capillaries, but the radial concentration gradient is assumed to be negligible. Third, the tracer concentration is assumed to have a homogeneous spatial distribution within the extravascular space, which is a compartment.

The adiabatic approximation assumes that the concentration of contrast material in the extravascular space is changing slowly (in a quasisteady state) relative to the rate of change in the intravascular space (capillaries) (21). The adiabatic approximation can be used to represent the impulse residue function, H(t), (21) as

(A1)

where t is time, T_c is the capillary transit time, and F is the liver blood flow; thus, F · T_c is liver blood volume according to the central volume principle (22). E is the extraction fraction (23) of contrast material, and V_c is the distribution volume of contrast material in the extravascular space. E relates to PS of liver capillaries by means of the following relationship (23):

(A2)

If the concentration (enhancement) of contrast material input to the liver, I(t), is known, then the measured liver parenchymal contrast enhancement curve, Q(t), can be calculated as the convolution of I(t) and H(t):

(A3)

where * is the convolution operator. The validity of Equation (A3) assumes that liver blood flow is constant and that Q(t) is linear with respect to I(t).

The liver is supplied by both the hepatic artery and the portal vein. Use of the hepatic artery as the supplier of contrast material to the liver is impractical because of its close proximity to the portal vein. The high concentration of contrast material in the portal vein will cause beam-hardening artifacts, which can affect the contrast enhancement curve generated from the hepatic artery (24). Also, the relatively small size of the hepatic artery may result in partial volume averaging and underestimation of contrast enhancement in the hepatic artery (25). To a close approximation, the enhancement of the aorta can represent that of the hepatic artery input. Therefore, I(t) can be expressed as

(A4)

where A(t) and V(t) are contrast enhancement values of the hepatic artery (ie, the aorta) and the portal vein, respectively, and (or HAF) is the fraction of liver blood flow contributed by the hepatic artery. The values of (Eq [A4]), F (Eq [A3]), and E, T_c, and V_c (Eq [A1]) are changed iteratively to achieve an optimum fit by (Eq [A3]) to the measured parenchymal contrast enhancement curve.

	ACKNOWLEDGMENTS

 
We thank Dr Tinsu Pan, PhD, from the University of Texas, M.D. Anderson Cancer Center, for helping design the scanning protocols.

	FOOTNOTES

 

Abbreviations: FDG = fluorine 18 fluorodeoxyglucose • HAF = hepatic arterial fraction • HBF = hepatic blood flow • HBV = hepatic blood volume • PS = capillary permeability surface area product • ROI = region of interest • SUV = standardized uptake value • T₀ = appearance time of contrast material in tissue relative to that in the aorta

See Material and Methods for pertinent disclosures.

Author contributions: Guarantor of integrity of entire study, T.Y.L.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; manuscript final version approval, all authors; literature research, E.E.S., T.Y.L.; experimental studies, all authors; statistical analysis, E.E.S., T.Y.L.; and manuscript editing, all authors

	References

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX References

 

1. Johnson PC. Autoregulation of blood flow. Circ Res 1986;59:483–495.[Free Full Text]
2. Jones RD, Berne RM. Evidence for a metabolic mechanism in autoregulation of blood flow in skeletal muscle. Circ Res 1965;17:540–554.[Abstract/Free Full Text]
3. Lehtio K, Oikonen V, Gronroos T, et al. Imaging of blood flow and hypoxia in head and neck cancer: initial evaluation with [(15)O]H(2)O and [(18)F]fluoroerythronitroimidazole PET. J Nucl Med 2001;42:1643–1652.[Abstract/Free Full Text]
4. Mankoff DA, Dunnwald LK, Gralow JR, et al. Blood flow and metabolism in locally advanced breast cancer: relationship to response to therapy. J Nucl Med 2002;43:500–509.[Abstract/Free Full Text]
5. McFalls EO, Murad B, Haspel HC, Marx D, Sikora J, Ward HB. Myocardial glucose uptake after dobutamine stress in chronic hibernating swine myocardium. J Nucl Cardiol 2003;10:385–394.[CrossRef][Medline]
6. Sokoloff L. Relationships among local functional activity, energy metabolism, and blood flow in the central nervous system. Fed Proc 1981;40:2311–2316.[Medline]
7. Kallinowski F, Schlenger KH, Kloes M, Stohrer M, Vaupel P. Tumor blood flow: the principal modulator of oxidative and glycolytic metabolism, and of the metabolic micromilieu of human tumor xenografts in vivo. Int J Cancer 1989;44:266–272.[Medline]
8. Brown RS, Fisher SJ, Wahl RL. Autoradiographic evaluation of the intra-tumoral distribution of 2-deoxy-D-glucose and monoclonal antibodies in xenografts of human ovarian adenocarcinoma. J Nucl Med 1993;34:75–82.[Abstract/Free Full Text]
9. Kubota R, Kubota K, Yamada S, Tada M, Ido T, Tamahashi N. Active and passive mechanisms of [fluorine-18] fluorodeoxyglucose uptake by proliferating and prenecrotic cancer cells in vivo: a microautoradiographic study. J Nucl Med 1994;35:1067–1075.[Abstract/Free Full Text]
10. Teicher BA. Hypoxia and drug resistance. Cancer Metastasis Rev 1994;13:139–168.[CrossRef][Medline]
11. Lee TY. Functional CT: physiological models. Trends Biotechnol 2002;20(12 suppl):S3–S10.
12. Brix G, Ziegler SI, Bellemann ME, et al. Quantification of [(18)F]FDG uptake in the normal liver using dynamic PET: impact and modeling of the dual hepatic blood supply. J Nucl Med 2001;42:1265–1273.[Abstract/Free Full Text]
13. Graham MM, Peterson LM, Hayward RM. Comparison of simplified quantitative analyses of FDG uptake. Nucl Med Biol 2000;27:647–655.[CrossRef][Medline]
14. Ackerman NB. The blood supply of experimental liver metastases. IV. Changes in vascularity with increasing tumor growth. Surgery 1974;75:589–596.
15. Miles KA, Hayball MP, Dixon AK. Functional images of hepatic perfusion obtained with dynamic CT. Radiology 1993;188:405–411.[Abstract/Free Full Text]
16. Van Beers BE, Leconte I, Materne R, Smith AM, Jamart J, Horsmans Y. Hepatic perfusion parameters in chronic liver disease: dynamic CT measurements correlated with disease severity. AJR Am J Roentgenol 2001;176:667–673.[Abstract/Free Full Text]
17. Richardson PD, Withrington PG. Liver blood flow. I. Intrinsic and nervous control of liver blood flow. Gastroenterology 1981;81:159–173.
18. Schenk WG Jr, McDonald JC, McDonald K, Drapanas T. Direct measurement of hepatic blood flow in surgical patients: with related observations on hepatic flow dynamics in experimental animals. Ann Surg 1962;156:463–471.[Medline]
19. Hamacher K, Coenen HH. Efficient routine production of the 18F-labelled amino acid O-2-18F fluoroethyl-L-tyrosine. Appl Radiat Isot 2002;57:853–856.[CrossRef][Medline]
20. Johnson JA, Wilson TA. A model for capillary exchange. Am J Physiol 1966;210:1299–1303.[Free Full Text]
21. St Lawrence KS, Lee TY. An adiabatic approximation to the tissue homogeneity model for water exchange in the brain. I. Theoretical derivation. J Cereb Blood Flow Metab 1998;18:1365–1377.[CrossRef][Medline]
22. Meier P, Zierler KL. On the theory of the indicator-dilution method for measurement of blood flow and volume. J Appl Physiol 1954;6:731–744.[Medline]
23. Crone C. The permeability of capillaries in various organs as determined by use of the ‘indicator diffusion’ method. Acta Physiol Scand 1963;58:292–305.[Medline]
24. Joseph PM, Ruth C. A method for simultaneous correction of spectrum hardening artifacts in CT images containing both bone and iodine. Med Phys 1997;24:1629–1634.[CrossRef][Medline]
25. Cenic A, Nabavi DG, Craen RA, Gelb AW, Lee TY. Dynamic CT measurement of cerebral blood flow: a validation study. AJNR Am J Neuroradiol 1999;20:63–73.[Abstract/Free Full Text]
26. Materne R, Van Beers BE, Smith AM, et al. Non-invasive quantification of liver perfusion with dynamic computed tomography and a dual-input one-compartmental model. Clin Sci (Lond) 2000;99:517–525.[Medline]
27. Fukumura D, Yuan F, Monsky WL, Chen Y, Jain RK. Effect of host microenvironment on the microcirculation of human colon adenocarcinoma. Am J Pathol 1997;151:679–688.[Abstract]
28. Folkman J. Tumor angiogenesis. In: Loitta LA, Mendelsohn J, Israel MA, eds. The molecular basis of cancer. Philadelphia, Pa: Saunders, 1995; 206–232.
29. Manz R, Otte J, Thews G, Vaupel P. Relationship between size and oxygenation status of malignant tumors. Adv Exp Med Biol 1983;159:391–398.[Medline]
30. Claffey KP, Brown LF, del Aguila LF, et al. Expression of vascular permeability factor/vascular endothelial growth factor by melanoma cells increases tumor growth, angiogenesis, and experimental metastasis. Cancer Res 1996;56:172–181.[Abstract/Free Full Text]
31. Weber G. Enzymology of cancer cells. II. N Engl J Med 1977;296:541–551.
32. Weber G. Enzymology of cancer cells. I. N Engl J Med 1977;296:486–492.
33. Monakhov NK, Neistadt EL, Shavlovskil MM, Shvartsman AL, Neifakh SA. Physicochemical properties and isoenzyme composition of hexokinase from normal and malignant human tissues. J Natl Cancer Inst 1978;61:27–34.[Medline]
34. Ward-Hartley KA, Jain RK. Effect of glucose and galactose on microcirculatory flow in normal and neoplastic tissues in rabbits. Cancer Res 1987;47:371–377.[Abstract/Free Full Text]
35. DiPette DJ, Ward-Hartley KA, Jain RK. Effect of glucose on systemic hemodynamics and blood flow rate in normal and tumor tissues in rats. Cancer Res 1986;46:6299–6304.[Abstract/Free Full Text]
36. Vaupel P, Thews G. Pathophysiological aspects of glucose uptake by the tumor tissue under various conditions of oxygen and glucose supply. Adv Exp Med Biol 1976;75:547–553.[Medline]
37. Fukuda K, Taniguchi H, Koh T, Kunishima S, Yamagishi H. Relationships between oxygen and glucose metabolism in human liver tumours: positron emission tomography using 15O and 18F-deoxyglucose. Nucl Med Commun 2004;25:577–583.[CrossRef][Medline]
38. Zasadny KR, Tatsumi M, Wahl RL. FDG metabolism and uptake versus blood flow in women with untreated primary breast cancers. Eur J Nucl Med Mol Imaging 2003;30:274–280.[Medline]
39. Iwata Y, Shiomi S, Sasaki N, et al. Clinical usefulness of positron emission tomography with fluorine-18-fluorodeoxyglucose in the diagnosis of liver tumors. Ann Nucl Med 2000;14:121–126.[Medline]
40. Cremerius U, Bares R, Weis J, et al. Fasting improves discrimination of grade 1 and atypical or malignant meningioma in FDG-PET. J Nucl Med 1997;38:26–30.[Abstract/Free Full Text]
41. Ziegler SI, Haberkorn U, Byrne H, et al. Measurement of liver blood flow using oxygen-15 labeled water and dynamic positron emission tomography: limitations of model description. Eur J Nucl Med 1996;23:169–177.[CrossRef][Medline]
42. Raleigh JA, Dewhirst MW, Thrall DE. Measuring tumor hypoxia. Semin Radiat Oncol 1996;6:37–45.[CrossRef][Medline]
43. Weidner N, Semple JP, Welch WR, Folkman J. Tumor angiogenesis and metastasis: correlation in invasive breast carcinoma. N Engl J Med 1991;324:1–8.[Abstract]

This article has been cited by other articles:

				V. Goh, S. Halligan, A. Gharpuray, D. Wellsted, J. Sundin, and C. I. Bartram Quantitative Assessment of Colorectal Cancer Tumor Vascular Parameters by Using Perfusion CT: Influence of Tumor Region of Interest Radiology, June 1, 2008; 247(3): 726 - 732. [Abstract] [Full Text] [PDF]

				Y-x He and Q-y Guo Clinical applications and advances of positron emission tomography with fluorine-18-fluorodeoxyglucose (18F-FDG) in the diagnosis of liver neoplasms Postgrad. Med. J., May 1, 2008; 84(991): 246 - 251. [Abstract] [Full Text] [PDF]

				J. van den Hoff, T.-Y. Lee, and E. Stewart Blood Flow Quantification with Permeable Contrast Agents: A Valid Technique? Radiology, June 1, 2007; 243(3): 909 - 910. [Full Text] [PDF]

This Article Abstract Figures Only Full Text (PDF) All Versions of this Article: 2393041382v1 239/3/740    most recent Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal