Reduction of Image Artifacts in Mice by Bladder Flushing with a Novel Double-Lumen Urethral Catheter

Introduction

Ideally a contrast agent should have a half-life such that it provides sufficient signal during imaging yet does not stay in the body for long periods after the image has been acquired. Unfortunately, images using filtered back-projection and contrast agents or radionuclides can be confounded by accumulation of material in the bladder. This is due to the assumption that the object is essentially static during image acquisition. Techniques that use projections to reconstruct images include information from all of the signal in the imaged volume. This includes the potentially huge signals from large concentrations of contrast agent in the bladder, which can mask the signal from nearby tissues of interest.

If the concentration of the contrast agent changes significantly both spatially and temporally, that is, accumulates in the bladder, then there is inconsistent information in the projections. Streak artifacts occur when the concentration in the bladder is near zero for projections early in the image acquisition procedure and increases in the later projections. This confounds imaging the bladder and surrounding tissues. In small-animal imaging, this includes leg-borne tumors.

It is possible to image a dynamic object when the change is predictable; for example, cardiac motion can be accounted for by gating. Similarly, contrast agent clearance can be corrected for when it changes uniformly in time (not spatially) by normalizing the projections to initial concentration of contrast agent.

As molecular imaging continues to increase in popularity, the need to identify a weak signal (tagged molecule) in the presence of a much larger signal (bladder) will increase. For clinical PET, Koyama et al. [1] used bladder flushing in human subjects to attenuate artifacts caused by accumulation of ¹⁸F-labeled fluorodeoxyglucose (¹⁸F-FDG) in the bladder, whereas Penney et al. [2] chose to use analytical methods to reduce bladder signal artifacts in clinical SPECT. The imaging of mouse models using a labeled drug in preclinical testing has recently gained in popularity. Images can provide distribution information without the necessity of animal sacrifice, organ dissection, and determination of the compound concentration in individual organs [3]. The temporal evolution of compound distribution is even more difficult. These difficulties are eliminated using sequential images.

Noncomputational techniques for eliminating bladder artifacts have not been applied in mice. Many microPET images of small animals involve some bladder accumulation, which may hinder accurate image reconstruction. In Ref. [4], accumulation of contrast agent in the bladder is clearly identified in the PET images and the authors mention that this could be a disadvantage if the region of interest is near the kidneys or bladder. Interestingly Moadel et al. [5] suggest that a Foley catheter could be used to continuously irrigate the bladder not for reducing bladder signal but to alleviate any possible toxicity of high, therapeutic doses of positron-emitting ¹⁸F-FDG (positherapy) in patients. Yet, they mention and show bladder activity in their mouse image.

Electron paramagnetic resonance (EPR) images are produced from projections obtained with fixed stepped gradients. The nontoxic triarylmethyl (trityl) spin probe, OX063, developed by Nycomed Innovations (now part of GE Healthcare, Malmo, Sweden) is useful as a narrow-linewidth spin probe for in vivo EPR imaging (EPRI). Physiological information, for example, oxygen concentration, is derived from the width of the spectral line of the trityl in a spectroscopic image, analogous to magnetic resonance spectroscopic imaging. One benefit of Nycomed's trityl spin probes is their high solubility in water. On the other hand, this contributes to the accumulation of spin probe in the bladder. In a mouse, a leg-borne tumor is often less than 1 cm from the bladder. Concentrated trityl in the bladder can create an overwhelming source of signal, often several orders of magnitude greater than that from the tumor. At such high concentrations, the spectrum of the spin probe is distorted by a process of self-broadening. In this situation, the image of the tumor region is reconstructed with a source of signal of considerably higher intensity and with a nonphysiological pO₂ (inferred from the spectral linewidth) in the bladder, which can be nearby. This unwanted signal confounds the image of linewidth broadening by oxygen of spin probe in the tumor (the desired measurement). Bladder signal artifacts have contaminated many of the EPR images taken in the past. Simple catheterization of the bladder is not always sufficient, as the catheter, without a balloon used with humans, can be easily obstructed by the bladder wall and the drainage pressure is low. Therefore, unassisted drainage of the bladder can be limited and variable.

A method for flushing is required that is suitable for repeated multimodality imaging of a mouse. The key difficulty with a mouse is the very small size of the animal and components of its micturition system. This may seem daunting to some investigators. Thus, many of the methods published for flushing a mouse bladder are invasive and incompatible with MRI or EPRI [6–9]. We found that simple catheterization of the bladder with polyethylene-10 (PE10) tubing was unsuitable, as it was often sharp or abrasive and often caused abrasions regardless of attempts to “polish” the ends. This was also noted by Shapiro et al. [10]. Because PE10 tubing alone is too flexible for catheterization, a blunted trocar in an IV catheter was used, as it provided more control during insertion without trauma to the tissue. Dorr [11] and Lundbeck et al. [12] have published noninvasive procedures for repeated catheterization of a mouse bladder, using a 22-gauge (G) IV catheter. Although their method is MRI compatible, it is intended for filling, followed by drainage of the bladder, not continual flushing. Therefore, we have developed a novel, double-lumen urethral catheter to flush accumulated contrast agent from the female mouse bladder. The data presented here were taken as a subset of a larger project where quantitative EPR oxygen images were obtained as part of a multimodality suite of images being developed to guide antivascular gene therapy. They demonstrate a procedure that can be accomplished with a modest amount of practice.

Materials and Methods

The double-lumen catheter was a small-diameter polyethylene tube placed inside a standard IV catheter with a silicone tubing extension. Flushing water was introduced through the small-diameter tube and bladder effluent exits through the IV catheter, then through the silicone tubing extension. The concentric tubes were placed inside the bladder, with the small-diameter tube extending beyond the catheter. The inner tubing was stretched before insertion, causing it to curl within the bladder, reducing the chance of impingement of the catheter against the bladder wall. The inner PE10 tubing also extended beyond the outer silicone tubing at the distal end. This allowed the bladder effluent to drip from the distal end of the silicone tubing into a collection vessel.

The portion of the large-diameter tube that was inserted into the bladder consisted of a standard 20 G IV catheter (1.1 mm i.d., Introcan Safety IV Catheter, B. Braun Medical Inc., Bethlehem, PA). With a blunted trocar, this catheter was inserted into the urethra. The entire Luer adapter was cut off the IV catheter. For small mice, a 24 G IV catheter may be used to gain entry into the bladder. After trimming the 24 G catheter, a 20 G catheter can then slide over the 24 G catheter. If the 24 G catheter was used as a guide, it is removed after inserting the 20 G catheter. Then a piece of flexible silicone tubing (1.01 mm i.d. × 2.16 mm outside diameter (OD), VWR, West Chester, PA) was stretched over the 20 G catheter, tight enough to avoid fluid leakage. The inner tube consisted of PE10 tubing (0.28 mm i.d. × 0.61 mm OD, Clay Adams INTRAMEDIC Polyethylene, BD, Franklin Lakes, NJ). Prior to bladder cannulation, a short (~2.5 cm) tube of PE100 (0.86 mm i.d. × 1.52 mm OD, Clay Adams INTRAMEDIC Polyethylene, BD) was forced into the effluent (distal) side of the outer silicone tubing. This overcame the stickiness of the silicone tubing, making it easy to thread the PE10 tubing into silicone tubing. It was also a convenient place to tape the distal end of the outer tubing to the rim of a collection reservoir. The PE10 tubing was threaded into the PE100/ silicone outer tubing and into the bladder. The PE10 was marked beforehand to indicate the length required to extend a predetermined distance (~1.5 to 2 cm) beyond the catheter tip. The distal end of the PE10 tubing was connected to a Harvard 22 syringe pump (Harvard Apparatus, Inc., Holliston, MA), supplying water at 10 to 20 mL/hr (depending on the apparent concentration of the effluent) to be infused into the bladder. The effluent from the bladder exited from the 20 G catheter and flowed down the silicone tube into a collection reservoir. The apparatus is shown schematically in Figure 1, without the collection vessel. EPR imaging was performed as described previously [13–15].

The flow rate of 15 mL/hr was comparable to the flow rates reported in repeated mouse cystometry [11,12]. Because each mouse was imaged three times, a flow rate that would not interfere with the bladder physiology was desired. Water was used rather than saline, as the trityl was administered as a concentrated salt solution (2.5 g/kg, 350 mM). Because the mouse cannot be moved during image acquisition, a 20 G catheter was used rather than a 22 G (smaller outside diameter) to limit leakage.

Typical imaging mouse protocol is as follows: Female athymic nude mice (average weight, 23 g; average age, 7 weeks old) had their right hind limb injected subcutaneously with androgen-independent PC3 human prostate cancer tumor cells. Approximately 2 weeks later, the tumor reached a mean diameter of 5–8 mm, and the mice were imaged using EPRI. Treatment (combined antiangiogenic adenoviral gene and radiation therapy) was given on day zero (the baseline) after imaging. Three and 16 days later, the mouse was again imaged with EPR, therefore requiring survival of the animal with minimal morbidity.

To show the effect of bladder flushing, a female C3H mouse (22.9 g, 8 weeks old) had its right hind limb injected subcutaneously with fibrosarcoma (FSa) cancer tumor cells. The mouse (with a tumor size approximately 1 mL) was sacrificed after two sequential images were taken with and without bladder flushing without moving the mouse. To reduce temporal effects, these images were 3-D (two spatial, one spectral) rather than the 4-D (three spatial, one spectral) images used with the athymic nude mice.

All mice were anesthetized during imaging with a mixture of ketamine HCl (50 mg/kg)/xylazine (2.1 mg/kg) via an intraperitoneal line, or 1.5% to 3% isoflurane to maintain a surgical plane of anesthesia. A standard 24 G IV catheter (0.7 mm i.d., Introcan Safety IV Catheter, B. Braun Medical Inc.) was inserted into the tail vein for the injection of contrast agent. All animal procedures were performed under protocols approved by the Institutional Animal Care and Use Committee. It should be emphasized that the catheterization procedure described is for female mice due to the shorter length, larger diameter, and the higher compliance of female mice urethrae.

Results

A total of 63 images were examined: 36 without bladder flushing and 27 using the current bladder flushing procedure. The images were grouped into three categories: (1) no artifacts and no signal from the bladder, (2) with signal from the bladder but no streak artifacts, and (3) large signal from the bladder with streak artifacts. Typically, the bladder signal (without streak artifacts) was located by using 3-D visualization software written in MATLAB (The MathWorks, Inc., Natick, MA).

Prior to the development of the double-lumen catheter, 42% of the images were spoiled by image artifacts directly attributed to bladder signal. Another 28% had enough contrast agent in the bladder to make it visible in the image but not cause noticeable artifacts. After implementing the double-lumen bladder catheter with flushing, 7% of these images contained significant bladder-induced image artifacts and 74% of these images had no bladder signal. The remaining 19% of the images taken using bladder flushing had a small bladder signal but no artifacts.

Figure 2A shows an example of an EPR image of a mouse bladder (denoted by the arrow) with streak artifacts projecting from it. The self-broadening of the trityl reduces the signal height per spin, but the large number of spins in the bladder generates a huge signal. The tumor signal is overwhelmed and cannot easily be viewed. However, as shown in Figure 2B, when bladder flushing is used properly, the bladder signal is substantially eliminated. Furthermore, the heterogeneity of the tumor is clearly demonstrated, without obstruction of artifacts from the bladder signal. Figure 2B is typical of the good images acquired with sufficient bladder flushing.

OPEN IN VIEWER

Figure 1.

Double-lumen bladder flushing catheter, with the modified 20 G catheter highlighted in pale green. Flushing fluid enters at the right into the PE10 and exits at the left (denoted by the blue arrows). Effluent flows into the 20-gauge catheter tip and exits the silicone tubing at the right (denoted by the green arrows).

OPEN IN VIEWER

Figure 2.

(A) Sagittal view of an EPR intensity map of a PC3-tumor-bearing leg with insufficient bladder flushing. The arrow denotes the signal from the bladder. Note the artifacts projecting from the bladder signal. (B) Sagittal view of an EPR intensity map of a PC3-tumor-bearing leg with sufficient bladder flushing. The field of view is chosen to show just the upper leg and tumor. Color bars are in arbitrary units.

To show the effect of bladder flushing in the same mouse (without moving the mouse), Figure 3 also demonstrates that physiological data are corrupted by artifacts from the bladder. Although Figure 3 does not have the extreme artifacts seen in Figure 2, the oxygen map is distorted enough to make it not useful.

Discussion and Conclusions

This work demonstrates that artifacts from the bladder signal have been successfully attenuated by using a relatively simple method for flushing the bladders of female mice. It should be emphasized that this technique is compatible with our multiple time point, multimodality imaging protocol, which is dependent on mouse survival in a noninjured state. Although this technique is presented in the context of using a magnetic resonance substrate, it is clearly applicable to any imaging study where bladder accumulation will corrupt nearby portions of the image. The portions of the anatomy likely to benefit from this include the pelvic region and, for a relatively full bladder, lower abdominal region.

OPEN IN VIEWER

Figure 3.

(A) Sagittal view of an EPR intensity map of an FSa-tumor-bearing leg with no bladder flushing. (B) Sagittal view of the same FSa-tumor-bearing leg with sufficient bladder flushing. Color bars are in arbitrary units. (C, D) pO₂ maps of the same images as in (A) and (B) with color bars in millimeters Hg. The field of view is chosen to show just the upper leg and tumor. The black arrows denote the signal from the bladder; the red arrows denote an artifact from the bladder. The green arrows show the approximate location of the tumor. Note the hypoxic region in (D) near the green arrow. The intensity image color bars are in arbitrary units.

At this point, the major limitation of the technique is that only female mice can be used. The geometry and anatomy of male mice have prevented successful bladder catheterization using this technique. Several male mice of similar size and weight to those of the female mice used were tried. We feel the catheter is too rigid to curve into the bladder. Yet PE10 alone is too flexible and simply bends rather than advancing into the bladder.

Further refinements may be desirable. For example, it is possible for the tip of the bladder catheter to impinge against the bladder wall and therefore to disrupt the flow out of the bladder. An important innovation in transurethral catheter development could be the use of perfusion holes. We have tried making holes in the side of the catheter to improve fluid exchange, even when the wall of the bladder occludes the tip. This is likely to be the cause of the streak artifacts in two of the images using bladder flushing (7% of the 27 new images). In addition, this would allow a lower flow rate, thus reducing the potential for overdistending the bladder. Unfortunately, we were unable to make the holes such that the surface remained smooth enough so as not to abrade the lining of the urethra. Discussions are ongoing with a company with experience making such small holes in this tubing size.

Some of the trityl spin probes are difficult to synthesize, and the bladder flushing catheter also provides a means to collect the spin probe for recycling. The rapid renal extraction of these valuable molecules makes a large fraction of them potentially available for recycling. Depending on the recovery fraction, collection of the entire bladder effluent could multiply the effective supply of these compounds.

Although not yet completely optimized, the double-lumen urethral catheter provides a substantial reduction in artifacts when compared to images taken without bladder flushing or with poor flushing.