| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Experimental Studies |
1 From the Department of Biomedical Engineering (J.E.C., A.R.S., K.W.F.), Comparative Pathology Graduate Group (R.E.P.), Animal Resource Services (S.M.G.), and Department of Surgical and Radiological Sciences, School of Veterinary Medicine (E.R.W.), University of California, Davis, 1021 Academic Surge, Davis, CA 95616. Received May 6, 2002; revision requested July 24, 2002; final revision received February 19, 2003; accepted March 10. Address correspondence to K.W.F. (e-mail: kwferrara@ucdavis.edu).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: Mammary adenocarcinoma cells were implanted into the subcutaneous tissues of the flank of 11 rats. Resultant tumors were imaged serially with contrast-enhanced US and compared with images of the rat kidney, a highly perfused normal organ. The US acquisition and processing methods yield images of perfused tumor regions and the times required to achieve 80% replenishment. Findings at contrast-enhanced computed tomography (CT) and light microscopy of hematoxylin-eosin–stained tumor tissue were compared. Paired Student t test was performed to compare the accuracy of US with that of histologic examination and CT in the detection of viable tumor regions.
RESULTS: Replenishment of the kidney cortex microvasculature requires 1–5 seconds compared with a replenishment time of 6–14 seconds in tumors. Over the time course of tumor growth, the mean perfusion time becomes progressively longer, and a wider range of perfusion times is detected. Comparison of findings at US, CT, and histologic examination suggested that all three methods yield correlated estimates of the percentage of viable perfused tumor cells. Results of the t test suggested that the viable tumor percentages observed at US are not significantly different from those observed at CT and histologic examination (US vs CT, P = .92; US vs histologic examination, P = .94).
CONCLUSION: Repeated measurements of microvascular flow rate can be accomplished in a rat animal model with a minimally invasive technique.
© RSNA, 2003
Index terms: Angiogenesis • Breast neoplasms, US, 00.12988 • Ultrasound (US), contrast media, 00.12988
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
Development of a large number of angiogenic and antiangiogenic therapies has created the need for techniques that noninvasively quantify vascular flow changes in response to therapy. Methods to quantify blood volume and flow in tumors are required to assess the efficacy of biologically active chemotherapeutic agents whose mechanisms are unknown in some cases and whose response can be slow and unpredictable in human drug trials.
As a result of the low flow rate (7) and heterogeneous vascular density, the mapping of blood flow in tumors has been problematic. With the addition of contrast agents, US is sensitive to capillary-sized vessels and very low flow rates while maintaining the ability to detect morphologic information from traditional B-mode imaging (8). US contrast agents are microbubbles surrounded by a shell that reduces the rate of static diffusion. With an acoustic pulse of sufficient pressure, these microbubbles can be disrupted into a set of small fragments that can then dissolve into the surrounding plasma. The time required for the contrast agent to replenish each small volume is then estimated. In our imaging strategy, a destruction and replenishment method was used that was described in 1998, when Wei et al proposed the method for the assessment of blood velocity in the myocardium (9). Continued research has been performed in the assessment of contrast agent replenishment after acute myocardial infarction (9–12) and in other organs (13,14).
By using a destruction-replenishment scheme (15), we aimed to determine the ability of contrast-enhanced US to assess replenishment time in a rat kidney and adenocarcinoma tumor model.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
Contrast agent replenishment was measured in a rat kidney and rat mammary adenocarcinoma tumor model. The tumor cell line (R3230AC, provided by Mark Dewhirst, PhD, DVM, Duke University) was surgically implanted (R.E.P.) into the left or right flank regions of 11 healthy Fischer 344 rats (weight range, 300–500 g). The rats were allowed food and water ad libitum. Before imaging, the tumors were allowed to grow until they reached a volume of 1 cm3 (approximately 28 days). Each rat was anesthetized with an inhalational anesthetic (Isoflurane; Halocarbon Products, River Edge, NJ) delivered at 2% concentration with oxygen through a nonrebreathing anesthetic delivery system. A 24-gauge 0.75-inch (1.9-cm) catheter (Insyte; Becton Dickinson, Sandy, Utah) was inserted into a tail vein for intravenous injection of a microbubble US contrast agent suspension (MP1950; Mallinckrodt, St Louis, Mo).
After the rat was anesthetized, it was placed on a heating pad to maintain body temperature and to minimize any temperature-induced changes in blood flow. Because the tumors were very close to the skin of the rat, imaging gel and a gel pad were placed on the tumor to act as a standoff so that the center of the tumor was positioned approximately 2 cm from the transducer surface. Eleven tumors were evaluated serially with the imaging protocol. Our initial sample size estimate was based on projections of the expected time for replenishment in the tumor and kidney. With worst-case estimates of 4.5 and 3.5 seconds, respectively, with a 10% SD in each measurement, 
of .01, and double-sided test with power of 0.9, we calculated a sample size of greater than 10. The age of the rats in this study ranged from 12 to 24 months. We used relatively large rats for this study (mean weight, 400 g) to improve intravenous catheter access for US contrast agent administration.
US Imaging
Blood volume and velocity estimation at contrast-enhanced US is accomplished by means of detection of microbubbles as they are transported via blood vessels into a tissue of interest. These microbubbles can be destroyed with a destructive imaging mode, and the inflow of new microbubbles can then be observed with a nondestructive imaging mode. The time required for the echo intensity to reach 80% of the original intensity is estimated as a measure of blood velocity from the characteristic exponential increase of the received signal as the contrast agent returns to this region. This is termed the "replenishment time" in the remainder of this article. The imaging process required three steps: destructive imaging of the contrast agent, nondestructive imaging of the contrast agent, and estimation of parameters. The destructive imaging mode, with high pressure and low-frequency insonation, and the nondestructive imaging mode, with phase-inversion subharmonic imaging, were implemented (Elegra; Siemens Medical Systems, Issaquah, Wash).
Destructive imaging pulses generate a peak negative transmission pressure of 2.2 MPa at 3 MHz (mechanical index, 1.3). Findings in previous experimental work (15) indicate that this destructive imaging pulse will destroy all microbubbles with radii of less than 3 µm with a single ultrasonic pulse. The resting radius of the contrast agent used for this investigation is approximately Rayleigh distributed, with a mean radius of 1.0 µm, with 99% of bubbles having a radius of less than 3 µm.
We developed a nondestructive imaging mode based on subharmonic oscillations generated by means of insonification of contrast agent microbubbles at a specific frequency and amplitude. Subharmonic imaging with transmission frequency of 6 MHz and reception with a narrow bandwidth centered at 3 MHz invoked a subharmonic response from bubbles with radii of 0.5–1.5 µm. The ratio of microbubble echo amplitude to tissue echo amplitude is higher with subharmonic imaging because US does not generate subharmonic oscillations in tissue. A train of pulses with a center frequency of 6 MHz and peak negative pressure of 800 kPa (mechanical index, 0.3) was used to interrogate the flowing contrast agent, which resulted in minimal microbubble destruction. The pulse sequence includes many unevenly spaced pulse pairs over 34 seconds, with each pair including a rarefaction-first pulse and a compression-first pulse.
Imaging began 28 days from implantation of the tumor and was performed at least once a week or until either the 48th day from implantation or the tumor became too large or ulcerated to continue observation of the animal. The 48-day upper boundary was established on the basis of development of ulceration around this time. From experience, we knew that most tumors will ulcerate by 50 days, and continuation of the study beyond the point where tumors ulcerate is difficult technically and involves subjecting the rats to unnecessary pain and suffering. Forty-eight days was chosen as the longest time point at which most tumors would have an intact skin surface. The tumors were measured (R.E.P., J.E.C.) in three perpendicular axes by means of imaging of the largest extent of each axis and measurement of the length with the caliper tool on the US scanner. Both investigators measured all tumors independently and agreed on the final result. Tumor volume was estimated from three perpendicular axes of the tumor by means of a prolate ellipsoid method: [(length x width x height)
/6].
Contrast agent microbubbles were injected continuously into the tail vein catheter (R.E.P.) at a rate of 10 mL/h by using a syringe pump. The contrast agent bubbles were diluted into saline at a concentration of 130 µL of contrast agent to 1 mL of saline. Approximately 2 minutes after the start of contrast agent injection, the contrast agent concentration approached a uniform distribution, and the imaging sequence was applied. The full imaging sequence required up to 34 seconds for each replenishment estimate; typically, images were obtained with two to three sequences for each imaging plane on the tumor. Replenishment imaging in the kidney was performed with a faster imaging sequence, which lasted 16 seconds, to adequately sample the rapid influx of contrast agent in the kidney. Immediately after completion of the imaging sequence, the syringe pump was stopped. The total dosage of contrast agent was approximately 100 µL for a typical experiment, and the total time of observation was usually less than 15 minutes.
Replenishment Time and Integrated Contrast Enhancement
The replenishment time of contrast agents that entered the sample volume was estimated on the basis of an increasing exponential signal model, where the parameter ß is the exponential time constant. The echo from the destructive pulse was not used in the estimation. The signals from the remaining pulse pairs were used. First, the two echoes in each pair were summed. The result of the first sum was subtracted from all subsequent sums to yield signal with a high signal-to-noise ratio for contrast agent to tissue. This summation and subtraction reduced the tissue echo amplitude compared with the contrast agent echoes. The resultant signal amplitude was set to zero in pixels in which the signal amplitude increased by less than a threshold value over the pulse train. The threshold was set to eliminate signals in the surrounding skin and muscle. The replenishment time (in seconds) was then estimated for this resultant train of signals, where t80% = ln(0.2)/ß. Pixels in which contrast agent echoes increased in a monotonic fashion were encoded with a color to indicate the replenishment time.
The integrated contrast agent enhancement can also be calculated over the entire transmitted pulse sequence. To calculate the integrated enhancement, echoes from tissue were rejected by means of the same schemes of subharmonic reception, summation of returns from phase-inverted pulses, and subtraction of one summed pulse pair from the remaining echoes. This integration created a map with several orders of dynamic range.
Contrast-enhanced CT
On the last day of US imaging, contrast-enhanced CT of the tumor was performed (R.E.P., E.R.W.). Contiguous 1.5-mm CT images were acquired through the entire tumor with an incremental computed tomography (CT) scanner (9800 Quick; GE Medical Systems, Milwaukee, Wis) with a small field of view and a normal filter. The CT images were evaluated, and the section that most closely resembled the US image was chosen (R.E.P.). An iodinated ionic intravenous contrast agent (Conray 400 [400 mg of iodine per milliliter]; Mallinckrodt) was administered as a 2.0-mL intravenous bolus injection. Serial static CT images were obtained at the chosen section every 2 seconds for 1 minute after the onset of contrast agent administration.
Histologic Examination
Tumors were excised on the last day of imaging and preserved in 10% formalin (R.E.P.). The tumor specimens were embedded in paraffin, and 5-µm-thick slices were cut and stained with hematoxylin-eosin to observe the morphologic features of the tumor, most notably the viable and necrotic tumor regions (S.M.G.). Slices were stained with factor VIII, and regional microvascular density was counted (S.M.G.). Slices that corresponded to the midline of the tumor were imaged with magnification of x100 and digitized for analysis of the regions of viable tissue.
Statistical Analysis
A grid (approximately 1 x 1 mm) was overlaid on the images generated at US, CT, and histologic examination to estimate the percentage of viable tumor. The spatial extent of viable tumor was quantified as the percentage of tumor that exhibited enhancement at US and CT and as the percentage of viable tumor seen at histologic examination (J.E.C., K.W.F.). Each of the investigators measured the viable tumor in all tumors independently and agreed on the final result.
Paired Student t tests were used to compare the spatial extent observed at US, CT, and histologic examination at the final time point. A value of P < .05 was considered to indicate a statistically significant difference. Paired t tests were appropriate because we were comparing the same measurement (percentage of viable tumor) assessed in one group of rats with two imaging methods (CT and US) with a standard measurement (histologic examination), with all data collected at the same time point (day 48).
The time required for replenishment was also estimated by means of histograms of this time estimate compiled for each imaging plane in the kidney and tumor studies (J.E.C., A.R.S.). In the kidney, replenishment times in the cortical and medullary regions were segregated spatially, and the mean and variance of the collection of pixels in each region were evaluated (K.W.F.). Histograms were generated (MATLAB; MathWorks, Natick, Mass) for both the kidney and tumor replenishment times.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
|
|
|
|
|
Images acquired during a serial tumor study suggest an increase in mean replenishment time and replenishment heterogeneity over time (Fig 5). The tumor volume increased over the 24-day period from a volume of 0.5 cm3 on day 28 to 4.6 cm3 on day 42. A necrotic region that formed in the center of the tumor over time appeared as a dark region where contrast agent echoes were not detected. The maximum replenishment time increased with tumor growth in this long-term study, with large red regions corresponding to very long replenishment times that became apparent on days 39 and 42. The histologic slice that corresponds to Figure 5, E, is shown in Figure 2, B.
Integrated Contrast Enhancement
The morphology and size of the region with an elevated value for integrated enhancement was compared with those of the region that contained viable tumor at histologic examination and contrast-enhanced CT (Fig 6). The CT image (Fig 6, A) was acquired on day 42 and is aligned with the sagittal US image (Fig 6, D). A set of four integrated-enhancement US images is shown, with each sagittal image corresponding to a plane that cycled from the ventral side to the dorsal side of the tumor (Fig 6, B–D). The corresponding histologic section for Figure 6, D, is provided in Figure 2, B. In the histologic image, the viable tissue appears purple and correlates with the midline US image in that viable tumor appears in the tumor periphery and in a small region in the center of the tumor (Fig 6, D). The morphology of the contrast-enhanced region and that of the region containing viable tumor cells are similar. Typically, a necrotic core is surrounded by a peripheral ring of viable tumor cells. The improved image resolution with US compared with that with CT is apparent where the CT image shows consistent enhancement in the periphery of the tumor while contrast-enhanced US images show highly heterogeneous enhancement throughout the four planes that were imaged. At the deepest margin of the tumor, US correctly demonstrated an inconsistent rim of viable tumor that was present in some histologic slices and not in others. The US images depict small discrete regions of perfused tissue in the center of the tumor that coincide with viable tumor observed in the histologic slice.
|
Comparison of US, CT, and Histologic Findings
The spatial extent of viable tumor was computed in images from 11 tumors by overlaying a 1 x 1-mm grid on each image (Fig 7). A paired Student t test was performed with measurements of spatial extent from each modality acquired at the final time point, and the P value that is reported is the significance of a two-tailed hypothesized difference in the paired data. The spatial extent of perfused tissue observed at US was not significantly different from that observed at CT (P = .92) or that observed in the histologic slices (P = .94). This result suggests that the enhanced regions depicted at contrast-enhanced US correspond well with those depicted at contrast-enhanced CT and with regions of viable tissue seen in histologic slices. Within viable regions, no correlation was observed between local vessel density and US enhancement, US replenishment time, or CT enhancement.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
Comparison of findings with multiple imaging modalities with those at histologic examination is challenging because the section and slice thicknesses for each method are different, and accurate registration of the imaging planes with those tissue thicknesses is difficult. The section thickness with US is on the order of 1 mm and that with CT is 1.5 mm, whereas the slice thickness at histologic examination is 5 µm. The substantially larger section thicknesses in US and CT may lead to averaging over the elevational dimension compared with that at histologic examination; however, subjective evaluation of the US and CT images and the histologic slices in the current study demonstrates the potential to obtain correlated views.
Results of in vivo experiments show that the imaging method used in the current study is capable of mapping replenishment times in a healthy rat kidney, with resultant values that agree with previous estimates of microvascular flow rates (13). Destruction-replenishment imaging, as used in the current study, sweeps the destructive beam across an entire frame before the estimation of replenishment time in the entire region. In the kidney cortex, contrast agents must reperfuse the major arteries and flow across a substantial distance to refill the entire frame. As a result, replenishment times in the cortex of the kidney range from 1 to 5 seconds, with all regions demonstrating a rapid increase in the echo amplitude after a variable delay. We expected that a 1-mm3 US sample volume that contained capillaries would require 1–3 seconds to refill with contrast agent in a region adjacent to a major blood vessel. A delay on the order of 2 seconds is expected in regions located near the medulla and thus more than 1 mm from the terminal branches of the renal artery (the arcuate arteries) in the renal cortex (13). Histograms of replenishment times in the kidney cortex are narrow and symmetric, with an SD on the order of 2 seconds.
Results for replenishment times in tumors support those in previous studies (6) with laser Doppler US in the window chamber model. Time intervals of 6–14 seconds were required for replenishment in the tumor study in the current study, and the histogram of replenishment times in a tumor plane was broad as a result of variably and poorly perfused tissue and abnormal vascular architecture and tortuosity. A trend of increasing replenishment time with decreasing percentage replenishment was noted for values higher than 30%. For very low percentages of replenishment, the flow map was dominated by flow in small arteries rather than that in capillaries, and the mean and breadth of the replenishment time histogram decreased.
Estimation of replenishment time is simple and yields a useful measure of perfusion but may be susceptible to errors in cases where the signal model of an increasing exponential is not sufficient. Estimators that jointly estimate signal delay and rise time may decrease this error and will be investigated in future work.
The integral of the contrast agent echoes over the pulse sequence also provides an indication of the integrated flow. Given that tumor blood flow is expected to vary spatially in both velocity and volume, it may be important to evaluate measures of both quantities. Integrated US contrast enhancement demonstrated a spatial variation of three orders of magnitude in the tumor, as expected for tumor flow. Viable tumor demonstrated increased integrated contrast enhancement at US compared with that of surrounding tissue as a result of increased vascular density. Areas of contrast enhancement in the US and CT images were comparable to histologic maps of viable tumor tissue, where the cell nucleus was stained purple.
Comparison of the spatial extent of increased integrated enhancement at US and CT with viable tumor seen at histologic examination demonstrates that these estimates are correlated, with US depicting regions of perfused tissue with higher resolution and specificity for the microvasculature than those with CT. Alternative ways to analyze these data would be by means of a two-way analysis of variance or a correlation analysis. Results with both of these methods led to the same conclusion. There is good gross agreement between the measures of viable tumor among the three modalities. We are developing analytic tools to separate the effects of flow and vessel permeability on CT maps.
A combination of factors resulted in the lack of statistical correlation between regions of high microvascular density and areas with rapid or intense contrast enhancement at US. It is difficult to select a histologic slice that corresponds directly to the US imaging plane. In addition, tissue processing and fixation alters the shape of a given histologic slice, which makes acquisition of a direct spatial representation of the living tissue almost impossible. An alternative method for correlating microvascular density with blood flow depicted at US must be considered. Comparison with a standard of reference such as fluorescent or radioactive microsphere-derived blood-flow measures must also be considered.
Practical applications: Although we focused in this study on the mean replenishment time and gross comparisons with histologic findings, the images of replenishment times and enhancement morphology produced with this technique provide substantial information. Future studies are needed with use of US to monitor the replenishment time and volume of viable tumor cells over a course of therapy as a measure of therapy success. In such studies, techniques to detect and quantify vascular remodeling produced by antiangiogenic therapies may benefit from the detailed two- and three-dimensional maps produced with this technique.
|
FOOTNOTES |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
