Implanted Device

Durham, North Carolina 27704, and Kenan Plastic Surgery Research Laboratories,. Duke University Medical Center, Box 3906, Durham, North Carolina 27710...
0 downloads 0 Views 182KB Size
Anal. Chem. 2002, 74, 4849-4854

Magnetic Resonance Imaging of a Tissue/ Implanted Device Biointerface Using in Vivo Microdialysis Sampling Julie A. Stenken,*,† William M. Reichert,‡ and Bruce Klitzman‡,§

Department of Chemistry, 130 Cogswell Laboratories, Rensselaer Polytechnic Institute, 110 8th Street, Troy, New York 12180, Department of Biomedical Engineering, Duke University, Box 90281, Durham, North Carolina 27704, and Kenan Plastic Surgery Research Laboratories, Duke University Medical Center, Box 3906, Durham, North Carolina 27710

Real-time in vivo images of magnetic resonance contrast agent diffusion from implanted microdialysis probes were obtained by magnetic resonance (MR) microscopy. A gadolinium-containing contrast agent (Gd-DTPA) was infused through microdialysis probes implanted into the subcutaneous space of male Sprague-Dawley rats. The infusion of Gd-DTPA alters the T1 relaxation time for water protons near the microdialysis probe, thus causing an increase in brightness around the probe. Steady state concentration profiles of Gd-DTPA around the microdialysis probe were attained within 10 min. The distance for the diffusion of Gd-DTPA away from the probe was calculated to be ∼1400 µm on the basis of an image intensity analysis. A 5-cm field of view was used with a 256 × 256 matrix, giving a voxel volume of 0.190 mm3 (195 µm × 195 µm × 5,000 µm). These experiments demonstrate the ability of magnetic resonance microscopy to obtain real-time images of Gd-DTPA diffusion around implanted microdialysis probes. This noninvasive technique may be useful for determining how fibrous encapsulation during long-term implantation may affect localized mass transport at a biointerface. A vast research effort has focused on developing glucose sensors as either warning sensors or feedback control sensors for proposed artificial pancreas devices. A serious problem with implanted in vivo analytical sensors is the sensitivity alterations of the sensor during prolonged implantation times. Declination in sensor sensitivity during prolonged in vivo implantation is believed to be due to physiological and pathological changes at the sensor/tissue biointerface. A significant pathology is the fibrotic capsule formation occurring at the biointerface during the chronic phase of the wound healing or foreign-body response.1 Although the biological processes occurring during the foreignbody response to implanted materials and devices have been

described in the literature for a long time,2 it is not known how these processes affect in vivo sensor performance.3 In particular, a loss of blood capillaries around the sensor device as a result of fibrotic capsule formation would be expected to cause a decrease in analyte supply.4 One obstacle to the development of rugged and reliable long-term sensing devices has been the inability to noninvasively monitor analyte diffusive transport in real time through the fibrotic capsule at the sensor/tissue biointerface. In this paper, an implanted microdialysis probe is used as a mimic of an implanted sensor. Microdialysis sampling has been suggested as a potential technique to use for glucose monitoring in diabetic humans.5 Unlike biosensors that are prone to multiple types of failure mechanisms,6 microdialysis probes have been shown to reliably function in human subjects for long periods of time.7 The microdialysis probe will be used for a localized delivery of a hydrophilic magnetic resonance (MR) contrast agent (GdDTPA) at the probe/tissue biointerface. Microdialysis sampling is a well-established in vivo method for obtaining protein-free samples from animals and humans.8,9 Standard microdialysis practice uses an artificial cerebrospinal fluid solution (e.g., Ringers) that matches the ionic strength and composition of the fluid external to the semipermeable dialysis membrane that is passed through the microdialysis probe. Diffusion of analytes that are small enough to cross the semipermeable membrane is bidirectional, allowing both localized recovery and delivery. It is well-accepted from theoretical and empirical investigations that analyte mass transport during microdialysis sampling is affected by diffusive and kinetic properties of the analyte within the tissue.10-12 Concentration profiles around (2) (3) (4) (5) (6) (7)

* Corresponding author. Phone: (518) 276-2045. Fax: (518) 276-4887. E-mail: [email protected]. † Rensselaer Polytechnic Institute. ‡ Department of Biomedical Engineering, Duke University. § Kenan Plastic Surgery Research Laboratories, Duke University Medical Center. (1) Woodward, S. C. Diabetes Care 1982, 5, 278-281. 10.1021/ac020234w CCC: $22.00 Published on Web 08/10/2002

© 2002 American Chemical Society

(8) (9) (10) (11) (12)

Anderson, J. M. Annu. Rev. Mater. Res. 2001, 31, 81-110. Wilson, G. S.; Hu, Y. Chem. Rev. 2000, 100, 2693-2704. Sieminski, A. L.; Gooch, K. J. Biomaterials 2000, 21, 2233-2241. Weintjes: K. J.; Vonk, P.; Vonk-Van Klein, Y.; Schoonen, A. J. M.; Kossen, N. W. Diabetes Care 1998 21, 1481-1488. Wisniewski, N.; Moussy, F.; Reichert, W. M. Fresenius’ J. Anal. Chem. 2000, 366, 611-621. Hullegie, L. M.; Lutgers, H. L.; Dullaart, R. P. F.; Sluiter, W. J.; Wientjes, K. J.; Schoonen, A. J. M.; Hoogenberg, K. Neth. J. Med. 2000, 57, 13-19. Elmquist, W. F., Sawchuk, R. J., Eds. Adv. Drug Delivery Rev. 2000, 45 (special issue), 123-307. Lunte, C. E., Ed. Anal. Chim. Acta, 1999, 379 (special issue), 227-370. Bungay P. M.; Morrison P. F.; Dedrick R. L. Life Sci. 1990, 46, 105-119. Smith, A. D.; Justice J. B. J. Neurosci. Methods 1994, 54, 75-82. Stenken, J. A. Anal. Chim. Acta 1999, 379, 337-358.

Analytical Chemistry, Vol. 74, No. 18, September 15, 2002 4849

a microdialysis probe are highly dependent upon the localized tissue properties at the probe/tissue biointerface. If the localized tissue properties are altered because of a long-term foreign body reaction, including edema and fibrous encapsulation, analyte concentration profiles from the microdialysis probe may be significantly altered. Magnetic resonance (MR) microscopy provides a noninvasive method to observe analyte diffusion to and from the sensor/tissue biointerface. Eccles and Callaghan introduced the concept of the NMR microscope and demonstrated its use for imaging plat stems with a resolution of 20 µm during a 36 min acquisition.13 Thus, MR microscopy is a technique based on the physics of magnetic resonance imaging that allows better resolution than typical MRI experiments.14 This technique has been used to examine a variety of histological changes in small animal tissues.15 MR microscopy has been previously used during microdialysis sampling to determine the extent of brain extracellular fluid volume changes after the infusion of various drugs through an implanted microdialysis probe.16 MR microscopy has also been used to determine the extent of fibrous encapsulation around silicon implants.17 Predictions of drug transport through the brain have been theoretically modeled by using the contrast agent Gd-DTPA coupled with MR imaging.18 A signal from the MR microscope greatly depends on the longitudinal relaxation rate (1/T1) and transverse relaxation rate (1/T2) of water protons. As 1/T1 increases, the signal increases, and as 1/T2 increases, the signal decreases. Different metals, such as Fe and lanthanide ions, will cause the T1 for water protons to decrease.19 If a T1-weighted sequence is used for imaging, the delivery of a T1-shortening agent to a tissue will cause the targeted signal to increase while the background signal remains the same. This causes an increase in the contrast between the regions with the T1-shortening agent and the background tissue.20 A commonly used contrast agent is gadolinium (III) (Gd3+). Several reviews are available describing the development and use of Gd-containing contrast agents.21,22 Because free Gd3+ is highly toxic, chelated forms of Gd, such as Gd-DTPA are commonly used in clinical settings. The relation between the concentration of GdDTPA in the tissue, [Gd]t, and the altered T1 values is described by eq 1,

1 1 ) + r1[Gd]t T1 T1,0

(1)

where T1,0 is the original T1 for the tissue, and r1 is the longitudinal (13) Eccles, C. D.; Callaghan P. T. J. Magn. Reson. 1986, 68, 393-398. (14) Callaghan, P. T. Principles of Nuclear Magnetic Resonance Microscopy; Oxford University Press: Oxford, 1993. (15) Zhou, X.; Johnson, G. A. In The Biomedical Engineering Handbook; Bronzino, J. D., Ed.; CRC Press: Boca Raton, FL, 1995; pp 1119-1133. (16) Benveniste, H.; Hedlund, L. W.; Johnson, G. A. Stroke 1992, 23, 746-754. (17) Qiu, H. H.; Hedlund L. W.; Neuman, M. R.; Edwards, C. R.; Black, R. D.; Cofer, G. P.; Johnson, G. A. IEEE Trans. Biomed. Eng. 1998, 45, 921927. (18) Kalyanasundaram, S.; Calhoun, V. D.; Leong, K. W. Am. J. Physiol. 1997, 273, R1810-R1821. (19) Koutcher, J. A.; Burt, C. T.; Lauffer, R. B.; Brady, T. J. J. Nucl. Med. 1984, 25, 506-513. (20) Haacke, E. M.; Brown, R. W.; Thompson, M. R.; Venkatesan, R. Magnetic Resonance Imaging. Physical Principles and Sequence Design; John Wiley & Sons: New York, 1999; p 367.

4850

Analytical Chemistry, Vol. 74, No. 18, September 15, 2002

relaxivity. Since T1 is related to the concentration of [Gd]t, tissues that have Gd-DTPA present will appear brighter than those that do have Gd-DTPA. [Gd]t can be found once the specific tissue value for r1 is determined and if appropriate MR pulse sequences are applied.23 Although microdialysis probes can function for long periods of time, there are concerns reported in the literature suggesting the foreign body response affects the long-term calibration of these devices.24 Having the capability to monitor the diffusion distance of a marker for capillary permeability (Gd-DTPA) in real time during the time-course of a long-term implantation would be advantageous in determining the extent of the foreign body response around a microdialysis probe as well as understanding how the foreign-body response affects implanted sensors. By knowing the alteration of analyte mass transport during and after the creation of a fibrous capsule, the ability to predict and, thus, account for calibration alterations to the sensor or microdialysis sampling device may be possible. In this paper, we describe a noninvasive method to monitor the real-time diffusive transport of an analyte from a microdialysis probe implanted into rat subcutis using MR microscopy. EXPERIMENTAL SECTION Chemicals. Magnevist (Berlex Laboratories, Wayne NJ), a brand of gadopentetate dimeglumine (Gd-DTPA), was used as the contrast agent. Ringers irrigation solution contained 147 mM NaCl, 4 mM KCl, and 2.2 mM CaCl2‚2H2O (Abbott Laboratories, North Chicago, IL). Microdialysis Procedures. Male Sprague-Dawley rats weighing between 330 and 350 g were used and were maintained on a 12-h-on/12-h-off light cycle. Animals were allowed free access to food and water prior to surgical implantation of the microdialysis probes. The Duke University IACUC committee approved all animal procedures. Rats were anesthetized with ketamine/xylazine (40-80 mg/kg; 5-10 mg/kg) prior to microdialysis probe insertion. Sterile surgical procedures were used during the implantation of the microdialysis probes. Polyethersulfone microdialysis probes (10 mm long, 0.5 mm o.d.) were purchased from CMA Microdialysis (North Chelmsford, MA). The microdialysis probes were sterilized by exposure to ethylene oxide, outgassed for at least 48 h, and inserted into the subcutaneous tissue on the back of the rat parallel to the spine. After insertion, flow through the probes was ensured. When the microdialysis probes were not being perfused, the inlet and outlet tubes were connected together to prevent fluid evaporation. Imaging procedures commenced the following day. Magnetic Resonance Microscopy. All imaging experiments were performed on a GE 2 T horizontal bore magnet with 18 G/cm shielded gradients and a 15-cm clear bore. A 5-cm field of view was used with a 256 × 256 matrix, giving a voxel length of 195 µm. The slice thickness was 5 mm, giving a total voxel volume of 0.190 mm3 (195 × 195 × 5000 µm). (21) Lauffer, R. B. Chem. Rev. 1987, 87, 901-927. (22) Caravan, P.; Ellison, J. J.; McMurry, T. J.; Lauffer, R. B. Chem. Rev. 1999, 99, 2293-2352. (23) Rozijn, T. H.; van der Sanden, B. P. J.; Heerschap, A.; Creyghton, J. H. N.; Bovee, W. M. M. J. Magn. Reson. Mater. Phys., Biol., Med. (MAGMA) 1999, 9, 65-71. (24) Wisniewski, N.; Klitzman, B.; Miller, B.; Reichert, W. M. J. Biomed. Mater. Res. 2001, 57, 513-521.

Phantom Studies. To determine the appropriate concentration of Gd-DTPA to be used in the animal studies, solutions containing 5000, 500, 50, and 5 µM Gd-DTPA in Ringers solution were placed into NMR tubes. The Gd-DTPA solutions in the NMR tubes were placed into a rubber stopper that was placed into a hollow piece of polycarbonate tubing and placed into the MR scanner. These concentrations were chosen to span a range of T1 values to allow for acquisition of a signal with the MR repetition time, TR. To determine the ability of the microdialysis probe to deliver Gd-DTPA to the space external to the microdialysis probe, microdialysis probes were placed into a hollow polycarbonate tube containing 3% (w/w%) agar. The probes were perfused with 5 mM Gd-DTPA at 1.0 µL/min. The real-time images of the Gd-DTPA diffusion from the microdialysis probe were acquired with an echo time (Te) of 5.24 ms and a repetition time of 300 ms. A slice thickness of 5 mm was used for phantom studies. Animal Imaging Experiments. The rats were placed into a specially designed holder to allow for precisely timed mechanical ventilation synchronized with the magnetic pulses.25 Rats were maintained on 1.5-2.0% isoflurane in oxygen during all imaging procedures. To improve the signal intensity around the area in which the microdialysis probe was implanted, a surface coil was placed over this area.26 During imaging procedures, 5 mM GdDTPA in Ringers solution was infused through the microdialysis probe at 1.0 µL/min with a microinfusion pump (CMA/100, CMA Microdialysis, North Chelmsford, MA). The metallic infusion pump was kept at a minimum distance of 10 feet from the front of the magnetic coil. To allow perfusion of the microdialysis probes from this safe distance, 5 m of FEP tubing was connected together. Both time series and multislice pulse sequences were performed. For animal imaging, Te was set to 5.24 ms and TR was set to 200 ms. Image data analysis was performed by using the software program NIH image. RESULTS AND DISCUSSION To obtain images of the developing diffusion profile of GdDTPA around the microdialysis probe in real time, it was important to use a concentration of Gd-DTPA that allowed an image to be obtained with visually observable brightness around the microdialysis probe. Therefore, it was necessary to perform a phantom experiment to determine the concentration of Gd-DTPA that maximizes T1 reduction without significant alteration in T2. Figure 1 shows the phantom images of NMR tubes containing Gd-DTPA in Ringer’s solution at two different TR values (repetition time). The intensity for each of these calibration standards is listed in Table 1. Table 1 and Figure 1 clearly show that the brightest phantom was obtained with the 5 mM Gd-DTPA solution across the range of TR values chosen, except for when TR was set to 1280 ms. In Figure 1, TR was set to 320 ms, which gave the brightest image for the different Gd-DTPA solutions (left image), whereas when TR was set to 40 ms, the images for 50 and 5 µM Gd-DTPA had low intensity and could not be easily seen. There has been extensive work performed with chelated Gd contrast agents to determine the appropriate bolus dose to give to obtain high quality (25) Qiu, H. H.; Cofer, G. P.; Hedlund, L. W.; Johnson, G. A. IEEE Trans. Biomed. Eng. 1997, 44, 1107-1113. (26) Banson, M. L.; Cofer, G. P.; Hedlund, L. W.; Johnson, G. A. Magn. Reson. Imaging, 1992, 10, 929-934.

Figure 1. Magnetic resonance microscopy phantom inverse image of different Gd-DTPA concentrations using a Te of 5.24 ms: (A) TR (ms), 320 ms; (B) TR, 40 ms. Solutions of Gd-DTPA are labeled as (A) 5 mM (B) 500 µM (C) 50 µM, and (D) 5 µM. The small circle next to tube “A” is a capillary tube of CuSO4 that was used as a place marker. Table 1. Image Intensity for Gd-DTPA Calibrationa TRb

Tec

5000 µM

500 µM

50 µM

5 µM

1280 640 320 160 80 40 20

5.24 5.24 5.24 5.24 5.24 5.24 5.24

77.78 79.76 89.29 88.89 90.08 90.87 85.32

86.51 71.83 50.00 30.16 18.65 13.49 9.92

40.08 23.41 13.89 5.56 2.78 1.19 0.00

29.37 15.08 8.73 4.37 1.19 0.40 0.00

a Arbitrary intensity units of the phantom. 100 is the brightest possible image, and 0 is the darkest possible image. b Repetition time (ms). c Echo time (ms).

Figure 2. Inverse MR image of microdialysis probes in 3% agar. Probes were perfused with 5 mM Gd-DTPA in Ringers solution. (A) Image of the probes inserted in agar before the perfusion fluid was turned on. (B) Image of the probes inserted in agar after the perfusion fluid (1.0 µL/min) was turned on. Probe 2 was damaged and leaked Gd-DTPA along the side of the plastic wall.

in vivo images.27 However, Gd-DTPA has not been used to image around microdialysis probes at a tissue biointerface. Therefore, it was vital to perform an investigation of the proper concentrations of Gd-DTPA needed around the probe to obtain images with adequate brightness. The calibration experiments clearly illustrate that a concentration between 500 µM and 5 mM of Gd-DTPA is needed to allow for adequate imaging times with a reasonable duty cycle and appropriate brightness around the microdialysis probe. Thus, a TR of 200-300 ms was applied during time series acquisition of Gd-DTPA diffusion from implanted microdialysis probes. There are several different commercially available microdialysis probes with variable molecular weight cutoffs (MWCO) that can be used for in vivo microdialysis sampling. In general, MWCO (27) Runge, V. M.; Clanton, J. A.; Lukehart, C. M.; James, A. E., Jr. AJR 1983, 141, 1209-1215.

Analytical Chemistry, Vol. 74, No. 18, September 15, 2002

4851

Figure 3. Inverse MR images of microdialysis probes implanted in rat subcutaneous tissue: (A) microdialysis probe after the perfusion fluid flow was turned off for more than 15 min and (B) microdialysis probe after the perfusion fluid containing 5 mM Gd-DTPA was turned on to allow flow for more than 10 min.

Figure 4. Enlarged MR images of the developing Gd-DTPA concentration profile around an implanted microdialysis probe at different time points. The voxel area is 195 × 195 µm.

ranges between 6000 and 100 000 for commercially available microdialysis probes. Gd-DTPA has a molecular weight of 547 Da and Magnevist is formulated as Gd-DTPA with a dimeglumine salt that has a molecular weight of 938 Da. Since analyte diffusion properties affect the extraction efficiency during microdialysis sampling procedures, it was important to pick a membrane type that would ensure maximum loss of Gd-DTPA to the tissue space. This was especially important, since during these experiments, there were no instruments that could be easily accessed to 4852

Analytical Chemistry, Vol. 74, No. 18, September 15, 2002

measure the loss of Gd-DTPA across the microdialysis probe, for example, via ICPMS.28 Since there was uncertainty in the amount of Gd-DTPA loss across the microdialysis probe to the tissue, we chose to use 10-mm probes with a 100 000 Da MWCO polyethersulfone (PES) membrane rather than the oftentimes more rugged 20 000 Da cutoff polycarbonate membrane. With adequate analytical instrumentation to measure the Gd-DTPA concentrations, the (28) Skelly Frame, E. M.; Uzgiris, E. E. Analyst 1998, 123, 675-679.

Figure 5. Axial MR image of an implanted microdialysis probe.

microdialysis sampling procedures could be tailored with perhaps lower perfusion fluid flow rates to enhance the local concentrations of Gd-DTPA and, thus, brightness around the probes. A preliminary concern with coupling microdialysis sampling to magnetic resonance microscopy was the use of connecting tubing that was 5 m long. This long inlet tubing was needed to keep the metallic syringe pumps at a safe distance from the magnet (> 10 feet). Of particular concern was the knowledge that PES membranes are known to occasionally exhibit ultrafiltration during microdialysis sampling. Having the process of Gd-DTPA delivery governed by ultrafiltration rather than diffusion through the microdialysis probe would cause difficulties in the interpretation of the Gd-DTPA concentration profiles around the microdialysis probe. Prior to imaging, several PES microdialysis probes were infused with 5 mM Gd-DTPA in Ringers using 5 m of FEP tubing as the inlet and a short (8 cm) outlet. In vitro experiments of probes immersed in Ringers solution at room temperature using a flow rate of 2.5 µL/min through two different probes with varying times gave an average flow rate of 2.38 µL/min. This suggested that ultrafiltration rates may be only 0.12 µL/min during the infusion of Gd-DTPA through PES microdialysis probes used in this study. Figure 2 shows the MR images of the microdialysis probe inserted into a 3% agar solution. Figure 2A shows the microdialysis probe prior to initiation of Gd-DTPA infusion. Prior to the start of the infusion, there is some brightness in and around the area of the microdialysis probe, because there is some Gd-DTPA that remains in the cannula from the initiation process to pump out air from the probe. The perfusion pump was turned on, and images were obtained every 2 min. This phantom experiment showed that

it is possible to monitor the change in Gd-DTPA released around the microdialysis probe in real time. It was also important to determine whether the Gd-DTPA diffuses far enough from the microdialysis probe to allow observation of the formation of a concentration gradient. This is important, because the resolution for the in vivo experiments was 195 µm. If the presence of blood capillaries caused the Gd-DTPA to diffuse a very short distance as a result of uptake into the capillaries, then gradual differences in brightness around the microdialysis probe may not be observed because of the large voxel size. Figure 3 shows images of a microdialysis probe implanted into the subcutaneous tissue of a rat with Gd-DTPA perfused through the probe. Figure 3b shows the developed concentration profile around the microdialysis probe. This concentration profile developed quickly and reached a steady state within 10 min. Figure 4 shows enlarged images of the implanted microdialysis probe for the concentration profile during a 14 min time span. Because the microdialysis probe has to be cleared with the Gd-DTPA solution, there is some brightness in the “no flow” image as a result of the residual Gd-DTPA that remains in the microdialysis cannula during the flushing procedure. Table 2 shows the data for the radius of the concentration profile obtained as a function of image time. Although eq 1 can be used to obtain the concentration of Gd-DTPA in each voxel with the appropriate MR pulse sequence, the experiments described here did not use these types of sequences. Figure 5 shows an axial view of the microdialysis probe implanted in the subcutaneous tissue of the rat. What is surprising about this image is the noncircular nature of the image intensity around the microdialysis probe. It is generally expected that during microdialysis sampling, concentration profiles around the Analytical Chemistry, Vol. 74, No. 18, September 15, 2002

4853

Table 2. Distance of Image Brightness around an Implanted Microdialysis Probea time (min)

total distance (µm)

2 4 6 8 10 12 14 16 18 20 22

437 552 552 1248 1523 1406 1406 1379 1379 1379 1379

a The distance measured is the diameter of brightness across the probe. The dialysis probe is known to be 500 µm.

probe would be completely circular as a result of the cylindrical geometry of the probe. However, the implantation into the subcutaneous space clearly shows that multiple tissue properties greatly affect the Gd-DTPA diffusion profile around the microdialysis probe. Because the subcutaneous space contains a large amount of fluid, it is not surprising that diffusion profiles are in the direction of this space. In this work, the voxel area was 195 µm × 195 µm. Decreased voxel areas would greatly improve the technique in terms of the error associated with measuring the distance of the concentration profiles. The image resolution that can be achieved using MR microscopy of a live animal is a function of both the number of signal-averaged scans as well as interference from the movement of the animal during respiration. Better images are often achieved with euthanized animals in which the technique is used to perform whole-body histological analyses with high numbers of scans, thus increasing the signal-to-noise ratio.29 (29) Johnson, G. A.; Benveniste, H.; Black, R. D.; Hedlund, L. W.; Maronpot, R. R.; Smith, B. R. Magn. Reson. Q. 1993, 9, 1-30.

4854

Analytical Chemistry, Vol. 74, No. 18, September 15, 2002

CONCLUSIONS MR microscopy coupled with infusion of a Gd-DTPA as a contrast agent via a microdialysis probe can be used to obtain noninvasive real-time images of a device/tissue biointerface. This unique tool can be used to determine differences in analyte mass transport between acute and chronically implanted microdialysis probes. Such information would be quite useful to researchers interested in developing means to promote angiogenesis at implanted sensor/tissue biointerfaces. By using MR microscopy to study the molecular interactions at the sensor/tissue biointerface, the development time for successful long-term implantable sensors may be shortened, because the efficacy of such a bioengineered approach could be validated. Furthermore, a quantitative understanding of alterations to the diffusive and mass transport properties of the fibrotic capsule after subcutaneous implantation is of significant importance for long-term microdialysis sampling and ultimately for realization of proposed integrated in vivo combination smart drug delivery/sensing devices. ACKNOWLEDGMENT NSF CAREER CHE-9984150 awarded to Julie A. Stenken, NIH DK/HL54932 awarded to William M. Reichert, and NIH/NCRR P41 RR05959 awarded to the Center for In Vivo Microscopy (CIVM) at Duke University Medical Center supported this work. Ted Wheeler of the CIVM is gratefully acknowledged for his help with animal experiments during image acquisition. Professor G. Allan Johnson, Director CIVM, is gratefully acknowledged for discussions about some aspects of the experimental design. Professor George Wilson, University of Kansas, is acknowledged for helpful discussions about this manuscript. J.A.S. acknowledges all the kind support personnel who helped make her short stay at Duke University tremendously productive. Received for review April 10, 2002. Accepted June 26, 2002. AC020234W