Magnetic Resonance in Medicine 49:249 –257 (2003) Renal and Systemic pH Imaging by Contrast-Enhanced MRI Natarajan Raghunand,1* Christine Howison,1 A. Dean Sherry,2,3 Shanrong Zhang,2 and Robert J. Gillies1 Perturbations of renal and systemic pH accompany diseases of the kidney, such as renal tubular acidosis, and the ability to image tissue pH would be helpful to assess the extent and severity of such conditions. A dual-contrast-agent strategy using two gadolinium agents, the pH-insensitive GdDOTP5ⴚ and the pH-sensitive GdDOTA-4AmP5ⴚ, has been developed to generate pH maps by MRI. The renal pharmacokinetics of the structurally dissimilar pH-insensitive contrast agents GdDTPA2ⴚ and GdDOTP5ⴚ were found to be similar. On that basis, and on the basis of similarity of structure and charge, the renal pharmacokinetics of GdDOTP5ⴚ and GdDOTA-4AmP5ⴚ were assumed to be identical. Dynamic T1-weighted images of mice were acquired for 1 hr each following boluses of GdDOTP5ⴚ and GdDOTA-4AmP5ⴚ. The time-varying apparent concentration of GdDOTP5ⴚ and the time-varying enhancement in longitudinal relaxation rate following GdDOTA-4AmP5ⴚ were calculated for each pixel and used to compute pH images of the kidneys and surrounding tissues. MRI pH maps of control mice show acidic regions corresponding to the renal papilla, calyx, and ureter. Pretreatment of mice with the carbonic anhydrase inhibitor acetazolamide resulted in systemic metabolic acidosis and accompanying urine alkalinization that was readily detected by this dual-contrast-agent approach. Magn Reson Med 49: 249 –257, 2003. © 2003 Wiley-Liss, Inc. Key words: gadolinium; kidney; relaxivity; acetazolamide; acidbase The pH of bodily fluids affects the organism in many ways, especially through its effects on cellular and plasma proteins. Maintenance of acid-base homeostasis is critical, and occurs at several levels. The most immediate and local response to an acid or alkali load is through chemical processes, including intracellular, extracellular, and bone buffers. Buffering capacity is limited, and active and passive physiologic responses are also required to maintain the acid-base balance. These physiologic processes can be at the cellular level, such as through feedback changes in metabolism, and at the systemic level, involving adaptive changes to the excretion of volatile acids by the lungs and fixed acids by the kidneys (1). In most tissues the buffering 1 Cancer Center Division, University of Arizona Health Sciences Center, Tucson, Arizona. 2 Department of Chemistry, University of Texas at Dallas, Richardson, Texas. 3 Rogers Magnetic Resonance Center, University of Texas Southwestern Medical Center, Dallas, Texas. Grant sponsor: American Cancer Society; Grant number: IRG7400124451224; Grant sponsor: NIH; Grant numbers: R21-CA84697; R01-CA77575; R24-CA83148. Presented in part at the meeting of Contrast Media Research, Capri, Italy, 2001. *Correspondence to: Natarajan Raghunand, University of Arizona Cancer Center, 1515 N. Campbell Ave., Tucson, AZ 85724-5024. E-mail: [email protected] Received 25 February 2002; revised 9 September 2002; accepted 12 September 2002. DOI 10.1002/mrm.10347 Published online in Wiley InterScience (www.interscience.wiley.com). © 2003 Wiley-Liss, Inc. capacities of intra- and extracellular fluid are roughly equal, but the approximately 1.5:1 greater volume of intracellular fluid compared to extracellular fluid results in a greater contribution of intracellular buffers to overall buffering in these tissues (2). Certain pathologies are associated with a perturbed pH homeostasis. For example, tumor interstitial fluid has a reduced buffering capacity compared to normal tissue, and this, in combination with poor perfusion and increased lactic acid secretion by tumors (3), is believed to result in an acidic extracellular pH in tumors (4). This situation is exacerbated by the lowered buffering capacity of bicarbonate at more acid pH. This acidic pH of tumors has numerous consequences, including resistance to radio- and chemotherapies (5). Pathologically altered renal physiology can also manifest with perturbations in both systemic and renal pH. Hereditary defects and acquired deficiencies in renal tubular ion transport systems can result in systemic metabolic acidosis or alkalosis (6,7). Therapies to correct these conditions include the use of alkalinizing or acidifying buffers. Novel gene therapies have also shown promise for the treatment of some of these diseases, but there are still problems with heterogeneous gene delivery to the target tissue, and loss of corrective genes from that tissue over time (8,9). Methodologies to image the spatial distribution of tissue pH would have considerable biomedical and clinical relevance in such cases because they would enable the noninvasive assessment of disease extent, progression, and response to therapy. Several methods have been proposed to measure tissue pH by MRS and MRI. Some exploit endogenous MR resonances, while others require the administration of exogenous agents. Intracellular pH in tissues can be estimated from the 31P MR resonance of inorganic phosphate (Pi) (3). The pH-sensitive 31P MR resonance of 3-aminopropylphosphate (3-APP) has been used by us and other researchers to measure extracellular pH of tumors and normal tissues in mice (4,5,10). Molecules with pH-sensitive 19 F and 1H resonances have also been reported. Ojugo et al. (10) reported that pH measurements with excellent signalto-noise ratio (SNR) per time period are possible with the extracellular 19F-containing pH probe, ZK-150471. They list the lack of potentially interfering endogenous 19F resonances, and the higher NMR and pH sensitivity of ZK150471 as advantages of the 19F probe compared to 3-APP. Both 3-APP and ZK-150471 are cell-impermeant and report only the extracellular pH, although 31P MRS of 3-APP offers the possibility of simultaneous measurement of intracellular pH from the Pi resonance. Mason and coworkers (11) reported that a fluorinated derivative of vitamin B6, 6-fluoropyridoxol, readily enters cells, and they used this single molecule to measure the intra- and extracellular pH in rodent tumors by 19F MRS. 249 250 The 1H nucleus offers the highest inherent sensitivity, and it is possible to image the spatial distribution of tissue pH in vivo by the use of probes with pH-sensitive 1H resonances. We and others have employed an exogenously administered imidazole, IEPA, for imaging breast (12) and brain (13) tumor pH in rodents by MRS imaging (MRSI) of the pH-sensitive resonance of the protons on the C-2 carbon in the imidazole ring. This resonance is in the 8 –9 ppm region, with few interfering resonances. Endogenous imidazoles, such as histidine, also have pH-sensitive 1H resonances in this region, but their concentrations are too low to exploit. However, Vermathen et al. (14) successfully measured brain pH in humans by localized 1 H MRS after oral administration of histidine. MRS and MRSI techniques for measuring pH suffer a disadvantage in sensitivity, and therefore spatial resolution, but have the advantage that the measurement is independent of the concentration of the pH probe. A different approach to exploiting endogenous resonances to measure tissue pH has been proposed by Van Zijl and colleagues (15). They point out that the region of the 1H MR spectrum ⬎ 5 ppm contains several very low and undetectable resonances from NH protons that are typically in fast exchange with solvent water. This exchange process renders these resonances nearly impossible to detect with pulse sequences that include a water presaturation pulse. However, using a pulse sequence that permits visualization of these resonances, Van Zijl et al. (15) demonstrated that it is possible to measure pH from the pH-sensitive rate of exchange with water protons of one or more of these protons. Ward and Balaban (16) demonstrated that a compound (such as 5,6-dihydrouracil) with multiple proton exchange sites, each with different pH dependencies, may be used to measure pH using a “ratiometric” method. pH-sensitive gadolinium complexes (17,18) and gadolinium-containing pH-sensitive polyion complexes (19) offer the possibility of imaging pH with a spatial resolution comparable to standard MRI. However, the enhancement observed in an image will be dependent on not only the pH, but also the local concentration of the agent. One approach to determine pH is to sequentially administer two contrast agents with identical tissue pharmacokinetics, one being insensitive to tissue pH and the other being pH-sensitive (20). The distribution of the pH-insensitive agent can be used to predict the concentration of the pHsensitive agent. Recently a gadolinium-based pH-sensitive contrast agent, GdDOTA-4AmP5⫺, was introduced by Sherry and coworkers (17). The H-bonding network created by protonation of the phosphonates on the side-arms of this complex provides a catalytic pathway for exchange of the bound water protons with protons of bulk water, making the longitudinal water proton relaxivity (r1) of this molecule pH-sensitive. In the present work we demonstrate computed pH images of kidneys and nearby tissues in mice following sequential injections of GdDOTA4AmP5⫺ and the pH-insensitive tetraphosphonate analog, GdDOTP5⫺. The pharmacokinetics of the pH-insensitive agent were monitored prior to and following injection of the GdDOTA-4AmP5⫺, to alert us to changes in animal physiology during the course of the experiment. When the pharmacokinetics of the two GdDOTP5⫺ boluses com- Raghunand et al. pared favorably, we reasoned that the intervening GdDOTA-4AmP5⫺ would have followed the same time-dependent distribution. METHODS In Vivo MRI Experiments All MRI experiments were performed on a Bruker Biospec® 4.7 T system with 14 G/cm self-shielded gradients. Female SCID and C3H mice weighing 19 –26 g were imaged using a 25-mm-ID Helmholtz coil (Doty Scientific, Columbia, SC). Mice were anesthetized by inhaled isoflurane (1.5%, rest O2 at 1 L/min), and cannulated at the tail-vein. To minimize breathing-related motion, the animals were taped down gently into a holder, which was then positioned inside the coil. Temperature was monitored during all in vivo MRI experiments using a rectal fluoroptic temperature probe (Luxtron Corporation, Santa Clara, CA), and was maintained at 37–37.5°C by a recirculating warm-water blanket. Two contrast agents were used in any given animal: GdDOTP5⫺ (Macrocyclics, Dallas, TX), and either GdDOTA-4AmP5⫺ (Macrocyclics) or GdDTPA2⫺ (Magnevist®; Berlex, Fairfield, NJ). The structures and molecular weights of these contrast agents are shown in Fig. 1. The tail-vein cannula was capped with a heparinized septum to permit repeated injections. The dead volume of the infusion line was approximately 0.1 mL. Contrast agent (0.025– 0.1 mmole/Kg, as a 25-mM or 50-mM solution) was administered and chased with 0.15 mL heparinized saline over 40 s via this catheter at the appropriate time during each experiment. The slow rate of injection was used partly to allow the animal to tolerate the hypervolemic bolus, and partly because the reproducibility of slow injections was better than that of rapid boluses. Scout images were collected to plan coronal slices through the kidneys of the mouse. Once the slice scheme was finalized, a fat-suppressed proton-density spin-echo (SE) image was obtained with the following parameters: recycle time ⫽ 8 s, echo time (TE) ⫽ 6 ms, number of averages ⫽ 2, field of view (FOV) ⫽ 3.5 cm, matrix ⫽ 128 ⫻ 128, slice thickness ⫽ 1–1.3 mm, and number of slices ⫽ 2. The contrast-enhanced portion of each pH imaging experiment consisted of three consecutive phases: fat-suppressed SE images were acquired every 30 – 40 s for 1 hr following 1) a bolus of GdDOTP5⫺, 2) a bolus of GdDOTA4AmP5⫺, and 3) a bolus of GdDOTP5⫺, with the following imaging parameters: recycle time ⫽ 80 ms, TE ⫽ 6 ms, number of averages ⫽ 3– 4 (other parameters as given above for the proton-density image). Tissue pH maps were calculated for three mice from the pharmacokinetics of GdDOTP5⫺ and GdDOTA-4AmP5⫺ and the pH-dependent relaxivity of GdDOTA-4AmP5⫺ measured in vitro at 37°C and 4.7 T. In separate experiments, the renal pharmacokinetics of the two pH-insensitive molecules GdDOTP5⫺ and GdDTPA2⫺ were compared in four mice. Acetazolamide is a carbonic anhydrase inhibitor that causes metabolic acidosis and concomitant urinary alkalinization by inhibiting resorption of filtered HCO3⫺ (21). Mice were administered 30 mg/Kg/day acetazolamide (Sigma, St. Louis, MO) once a day orally for 7 days prior to pH Imaging Using GdDOTA-4AmP5⫺ 251 FIG. 1. a: GdDTPA2⫺, mol. wt. 545.6, assuming no protonation at pH 7.4. b: GdDOTP5⫺, mol. wt. 699.5, assuming 2 H⫹ bound at pH 7.4. c: GdDOTA-4AmP5⫺, mol. wt. 927.7, assuming 2 H⫹ bound at pH 7.4. imaging (21,22). The dose on the day of the imaging experiment was administered 1 hr prior to induction of anesthesia in the animal. All experiments were carried out in compliance with the guidelines of the University of Arizona Institutional Animal Care and Use Committee. All data were processed using programs written in-house in Interactive Data Language (IDL; Research Systems, Boulder, CO). In Vitro Relaxivity Measurements GdDTPA2⫺ and GdDOTP5⫺ solutions were made in 10 different concentrations of 0 –5 mM in phosphate-buffered saline (PBS), pH 7, loaded in 2-mL microcentrifuge tubes, and the water-proton T1 values were measured at 37°C and 4.7 T. Similarly, the water-proton relaxivity of GdDOTA-4AmP5⫺ was measured in PBS at several pH values between 5.75– 8.0. T1 measurements were made using a 12-segment, high-pass, 70-mm-ID birdcage coil. Sample temperature, maintained at 37°C by blowing warm air through glass wool surrounding the sample tray, was monitored using a fluoroptic temperature probe. Longitudinal relaxation times were measured by SE imaging using 20 recovery times of 40 – 8000 ms, with a TE of 6 ms. Small molecules such as these are excreted by glomerular filtration, and most of the contrast agent detected in kidney by MRI is expected to reside in the postglomerular filtrate. The filtrate normally has a low protein content due to the low permeability of the glomerular capillary wall to large molecules (23). Thus, PBS rather than serum or other protein-containing solution was chosen as the medium for determination of in vitro relaxivities. isolated compartments: 1) the filtrate, containing a high concentration of the contrast agent; and 2) reabsorbed water, containing essentially no magnetopharmaceutical. If water exchange between these two compartments is restricted, it would explain the reduced apparent relaxivity of GdDTPA2⫺ observed in the kidney in vivo. Landis et al. (26) demonstrated that the assumption of fast water exchange between intra- and extracellular compartments is invalid in rat muscle for concentrations of GdDTPA2⫺ greater than approximately 0.1 mM. This leads to an underestimation of the extracellular concentration of the contrast agent in typical dynamic MRI experiments, when relaxivities measured in vitro in saline or other cell-free solutions are directly applied to in vivo signal enhancement data. Consider the case in which water proton exchange between the reabsorbed water (compartment A) and filtrate (compartment B) in the kidney is near the slow-exchange limit. The Boltzmann equilibrium magnetizations (or proton-density signal intensities) in the two compartments are SA and SB, the corresponding precontrast longitudinal relaxation times are TA and TB, and the recycle time for SE imaging is TR. Neglecting T2-related signal decay (TE ⫽ 6 ms), and also assuming fast water mixing within each compartment, the net instantaneous magnetization (or precontrast signal intensity) can be written as: S 1 ⫽ S A 关1 ⫺ exp共 ⫺ TR /TA 兲兴 ⫹ SB 关1 ⫺ exp共 ⫺ TR /TB 兲兴. [1] Provided TR ⬍⬍ TA and TR ⬍⬍ TB, the exponential terms in the above equation can be approximated with Taylor series expansions to only the first-order terms: Effects of Compartmentation and Slow Water Exchange Tissue structure and physiology can alter the in vivo relaxivities of contrast agents. For example, GdDTPA2⫺ relaxivities in rat kidneys in vivo are substantially lower in the cortex and medulla as compared to in vitro relaxivities measured in saline, and to in vivo relaxivities measured in other tissues (24,25). This is possibly due to compartmentalization of the contrast agent molecules in the kidney and slow rates of water exchange between the compartments. Shuter et al. (24) noted that the kidneys create two S 1 ⬇ S A 䡠 T R/T A ⫹ S B 䡠 T R/T B. [2] After administration of GdDOTP5⫺, assuming that the contrast agent is only in compartment B, the postcontrast signal intensity can be approximated as: S 2 ⬇ S A 䡠 T R/T A ⫹ S B 䡠 T R ⫻ 关1 ⫹ r DOTP 䡠 T B 䡠 关GdDOTP 5⫺兴兴/T B [3] 252 Raghunand et al. where rDOTP is the longitudinal relaxivity of GdDOTP5⫺ in compartment B, which we assume to be the same as that measured in vitro in PBS. The above Taylor series approximation to the first-order terms only holds for regions in which TR is much less than the postcontrast longitudinal relaxation time TB/(1 ⫹ rDOTP 䡠 TB 䡠 [GdDOTP5⫺]), and may not be valid in regions with very high concentrations of contrast agent (such as the ureters). A short recycle time of 80 ms was used in order to maximize the validity of this approximation. Combining Eqs. [2] and [3] yields: 关GdDOTP 5⫺兴 ⬇ 共S 2 ⫺ S 1兲/共S B 䡠 T R 䡠 r DOTP兲. [4] As the compartment-specific SB cannot be measured, we have instead used (SA ⫹ SB), the sum of the magnetization in both compartments, obtained from the precontrast proton-density image. Thus, compartmentation of the contrast agent results in an underestimation of the concentration of the agent in compartment B by the factor SB/(SA ⫹ SB). It must be noted that in these experiments S2 and [GdDOTP5⫺] are functions of time, as the agent is cleared from the animal. One hour after administration of GdDOTP5⫺ an identical bolus of GdDOTA-4AmP5⫺ was administered to the animal. S3 represents the time-varying signal intensity following administration of GdDOTA-4AmP5⫺, and includes enhancement due to both residual GdDOTP5⫺ and fresh GdDOTA-4AmP5⫺. The combined concentrations of residual GdDOTP5⫺ and fresh GdDOTA-4AmP5⫺ are assumed to remain within the linear regime for dependence of signal enhancement on the concentrations of both agents. If the solution milieu in the filtrate (compartment B) is similar to saline, then it is worth noting that the signal enhancement provided by GdDOTP5⫺ and GdDOTA-4AmP5⫺ in PBS was linear up to 4 mM with our SE imaging parameters (vide infra). The concentration of contrast agent probably does not exceed 4 mM in the cortex or medulla for any significant length of time. However, it is probably in excess of 4 mM in the ureters (and bladder), as it gets concentrated at longer times postinjection. For the other regions the following expression can be written for the observed relaxivity of GdDOTA-4AmP5⫺, r4AmP: r 4AmP ⬇ S3 ⫺ S1 S B 䡠 T R 䡠 关GdDOTA ⫺ 4AmP 5⫺兴 ⫺ r DOTP 䡠 ⬇ 关GdDOTP 5⫺,residual兴 关GdDOTA ⫺ 4AmP 5⫺兴 from the data collected 30 – 60 min post-GdDOTP5⫺, and extrapolated for calculating [GdDOTP5⫺, residual] in each pixel at each time-point post-GdDOTA-4AmP5⫺. Let us consider the two terms on the right side of Eq. [5]. The underestimation of [GdDOTP5⫺] by use of (SA ⫹ SB) in place of SB in Eq. [4] results in overestimation of the first term. But we have used (SA ⫹ SB) in place of SB here too, and this cancels out the error in the first term. In the second term, both [GdDOTP5⫺, residual] in the numerator and [GdDOTP5⫺] in the denominator are underestimated by the same factor, and these errors cancel out as well. Thus, to a first approximation, we can use this dual-contrast-agent strategy in conjunction with T1-weighted imaging to compute the in vivo longitudinal water proton relaxivity of GdDOTA-4AmP5⫺, even in the presence of compartmentation of contrast agent and slow water exchange between the compartments. RESULTS AND DISCUSSION In Vitro Calibrations The longitudinal water proton relaxivity r1 of each contrast agent was calculated from a fit to the following equation: 1/T 1 ⫽ 1/T 10 ⫹ r 1 • 关Gd兴 where [Gd] is the concentration of the contrast agent, and T10 and T1 are the longitudinal relaxation times of water protons in the absence and presence of contrast agent, respectively. The relaxivity of GdDTPA2⫺ in PBS at 37°C and 4.7 T was measured to be 3.2 ⫾ 0.27 mM⫺1s⫺1, which is in good agreement (mutatis mutandis) with the values reported by Landis et al. (26) for GdDTPA2⫺ in saline at 37°C at 4.0 T and by Rozijn et al. (27) for GdDTPA2⫺ in water at 37°C and 6.3 T. The relaxivity of GdDOTP5⫺ in PBS at 37°C and 4.7 T was 3.0 ⫾ 0.34 mM⫺1s⫺1. Aime et al. (28) reported a value of ⬇3.3 mM⫺1s⫺1 for GdDOTP5⫺ at 60 MHz and 35°C. For GdDTPA2⫺, GdDOTP5⫺, and GdDOTA-4AmP5⫺ at all the pH values tested, signal enhancement (with TE ⫽ 6 ms) varied linearly with concentration up to 4 mM (r2 ⬎ 0.97 in all cases). Figure 2 shows the pH-dependent variation in water proton relaxivity (r4AmP) of GdDOTA-4AmP5⫺ at 37°C and 4.7 T. The data between pH 5.75– 8.0 can be described by the following polynomial function (r2 ⬎ 0.99): pH ⫽ 1.62 ⫹ 4.50•共r 4AmP兲 ⫺ 0.78•共r 4AmP兲 2. 关GdDOTP 5⫺,residual兴 S3 ⫺ S1 . 5⫺ ⫺ r DOTP 䡠 S B 䡠 T R 䡠 关GdDOTP 兴 关GdDOTP 5⫺兴 [5] The second portion of Eq. [5] depicts our assumption that the pharmacokinetics of GdDOTA-4AmP5⫺ will be essentially identical to that of GdDOTP5⫺, given their similar molecular structure and charge. S3 and [GdDOTP5⫺] are functions of time in this dynamic experiment, and the above equations are written for a given time-point. A biexponential function approximating the time-dependent decay of signal intensity was fitted on a pixel-by-pixel basis [6] [7] There is an inflection point near pH 5.75, and this equation is only applicable for computing the pH in samples within the pH range 5.75– 8.0. Unambiguous assignment of pH to a measured in vivo relaxivity can be made only in samples with pH values that fall strictly within any one monotonic region of the curve. For example, in our calculations a region at pH 5.25 would be misassigned to ⬇pH 6.48 on the other side of the inflection point. Renal Pharmacokinetics: GdDTPA2⫺ vs. GdDOTP5⫺ We compared the renal pharmacokinetics of GdDTPA2⫺ and GdDOTP5⫺ in four mice. The data from one animal are pH Imaging Using GdDOTA-4AmP5⫺ FIG. 2. The pH dependence of the water-proton longitudinal relaxivity of GdDOTA-4AmP5⫺ (r4AmP) in PBS at 37°C, 4.7 T. The solid line represents a curve-fit using a second-order polynomial function. In in vivo experiments pH was back-calculated from the calculated in vivo relaxivity of GdDOTA-4AmP5⫺ using the fitted relation pH ⫽ 1.62 ⫹ 4.5(r4AmP) – 0.78 (r4AmP)2 (r2 ⬎ 0.99). 253 shown in Fig. 3. T1-weighted SE images were collected dynamically for 1 hr following a bolus of GdDTPA2⫺ (0.05 mmole/Kg) using the parameters described above. A bolus of GdDOTP5⫺ (0.05 mmole/Kg) was then administered and images were collected for another hour. Figure 3a depicts the apparent concentration of GdDTPA2⫺ calculated at each time-point from three regions-of-interest (ROIs) in the kidneys, while Fig. 3b shows the apparent concentration of GdDOTP5⫺ calculated at each time-point for the same ROIs. As explained above, residual enhancement from the first (GdDTPA2⫺) bolus was estimated by extrapolation using a biexponential fitted function, and subtracted out prior to calculation of the apparent concentration of GdDOTP5⫺ at each time-point following the second bolus. The pharmacokinetics of GdDTPA2⫺ and GdDOTP5⫺ appear to be roughly comparable in the renal cortex and medulla, within the noise in the data in Fig. 3a and b. Some divergence in the apparent pharmacokinetics of the two molecules is visible, especially at the early timepoints, in the calyx-ureter ROI. The ROI data in Fig. 3a and b have an unavoidable component of noise arising from physiologic variability. The noise is more severe when the FIG. 3. Representative comparison of the renal pharmacokinetics of GdDTPA2⫺ and GdDOTP5⫺ in a female C3H/hen mouse. Data are averages from ROIs drawn by hand over the cortex (solid line), medulla (dotted), and calyx-ureter (dashed) in both kidneys and both imaging slices. Images were acquired for 1 hr each following first a bolus of GdDTPA2⫺ (0.05 mmole/Kg) and then a bolus of GdDOTP5⫺ (0.05 mmole/Kg). a and c: Data acquired following the GdDTPA2⫺ bolus. b and d: Data collected following the GdDOTP5⫺ bolus. The ordinate in panels a and b is the apparent concentration of the contrast agent calculated at each time-point. The ordinate in panels c and d is the AUC/time concentration calculated as per Eq. [8]. The signal enhancement following the GdDTPA2⫺ bolus was converted to concentration by using r1 ⫽ 3.2 mM⫺1 s⫺1. Residual enhancement from the GdDTPA2⫺ bolus was estimated by extrapolation and subtracted prior to calculation of the apparent concentration of GdDOTP5⫺ following the second bolus. 254 Raghunand et al. Table 1 Renal Pharmacokinetics of GdDTPA2⫺ and GdDOTP5⫺ ROI C3H/ hena Cortex Medulla Ureter C3H/ hena Cortex Medulla Ureter SCIDa Cortex Medulla Ureter SCIDa Cortex Medulla Ureter AUC/time at 30 min, mM 1st bolus AUC/time at 30 min, mM 2nd bolus Ratio bolus 2/bolus1 0.41 0.88 1.28 0.37 0.80 1.05 0.90 0.91 0.82 0.89 1.25 1.63 0.95 1.32 1.51 1.07 1.06 0.93 0.32 0.58 0.87 0.29 0.53 0.72 0.91 0.91 0.83 0.41 0.56 0.86 0.44 0.62 0.79 1.07 1.11 0.92 Renal Pharmacokinetics: GdDOTP5⫺ vs. GdDOTA-4AmP5⫺ a Bolus sequence for the C3H/hen mice was GdDTPA2⫺ followed by GdDOTP5⫺, and vice-versa for the SCID mice. data are examined on a pixel-by-pixel basis, making reliable computations of pharmacokinetics from these data challenging. Another complication is that medullary pixels contain both proximal and distal tubules, with the concentration of the contrast agent varying along the length of the nephron. To reduce these artifacts, an areaunder-the-curve approach was used to smooth the timevarying contrast agent concentration, C(t), in each pixel. The area-under-the-curve (AUC) for each pixel at each time-point was divided by the time postinjection (t) to yield AUC/time: AUC/time ⫽ 冕 t C共t兲dt 0 t . 0.87 ⫾ 0.05 in the calyx-ureter ROI (Table 1). The apparent AUC/time concentration of contrast agent in the calyxureter ROI was always higher 30 min following the first bolus of contrast agent than 30 min following the second bolus. The average concentration of contrast agent is higher in the calyx and ureters than in other regions of the kidney. It is possible that high total (residual ⫹ fresh) contrast agent concentration in the calyx-ureter ROI following the second bolus results in a significant shortening of T2 in these regions, with a reduction in the signal enhancement in our T1-weighted SE images. This would lead to the computation of a decreased apparent concentration of contrast agent following the second bolus. Hence, the pharmacokinetics were highly consistent in the cortex and medulla, and less so in the calyx-ureter. [8] Implicit in this manipulation is the assumption that signal enhancement maintains a linear dependence on contrast agent concentration in all pixels at all times. As discussed above, this assumption may fail in regions with very high contrast agent concentrations, such as the calyx and ureters. The data in Fig. 3a and b are replotted in Fig. 3c and d as AUC/time vs. time postinjection. In this example, agreement between the two pharmacokinetic curves was within 10% at 30 min postinjection in the cortex (0.89 mM vs. 0.95 mM), the medulla (1.25 mM vs. 1.32 mM), and the calyx-ureter (1.63 mM vs. 1.51 mM). Data from two different strains of mice, C3H/hen and SCID, are shown in Table 1. The sequence of contrast agent administration was GdDTPA2⫺ followed by GdDOTP5⫺ in the C3H mice, and vice versa in the SCID mice. Over all mice, the ratios of the AUC/time concentration of contrast agent at 30 min postinjection of the second bolus to the first bolus were 0.99 ⫾ 0.08 in the cortex, 1.0 ⫾ 0.09 in the medulla, and GdDOTP5⫺ and GdDOTA-4AmP5⫺ are arguably more similar to one another than are GdDTPA2⫺ and GdDOTP5⫺ (Fig. 1). Therefore, given the similarity in renal pharmacokinetics of the latter pair of molecules (Fig. 3 and Table 1), we made the assumption that the MRI pharmacokinetics of GdDOTP5⫺ is an adequate surrogate for the pharmacokinetics of GdDOTA-4AmP5⫺ in a given mouse. We also compared the uptake kinetics of GdDOTP5⫺ and GdDOTA4AmP5⫺, as well as the reproducibility of pharmacokinetics following two identical boluses of GdDOTP5⫺, in three mice. T1-weighted SE images were collected dynamically for 1 hr each following first a bolus of GdDOTP5⫺ (0.05 mmole/Kg), then a bolus of GdDOTA4-AmP5⫺ (0.05 mmole/Kg), and finally a second bolus of GdDOTP5⫺ (0.05 mmole/Kg). Representative data from one experiment are shown in Fig. 4a– c. The AUC/time apparent concentrations of GdDOTP5⫺ in three ROIs following the first and last bolus are plotted in Fig. 4a and c, respectively. The AUC/time enhancement in apparent relaxation rate following the middle bolus of GdDOTA-4AmP5⫺ is plotted for the same three ROIs in Fig. 4b. In the data plotted in Fig. 4b and c, residual enhancement from previous bolus(es) was accounted for as explained above. In Fig. 4a and c we compare the reproducibility of the pharmacokinetics of GdDOTP5⫺ in an animal. In this example, agreement between the two boluses was good at 30 min postinjection in the cortex (0.80 mM vs. 0.72 mM) and the medulla (0.89 mM vs. 0.98 mM), but not in the calyx-ureter (2.66 mM vs. 2.27 mM). Since the abscissa in Fig. 4b is different from those in Fig. 4a and c, we compared the pharmacokinetics of the three boluses on the basis of “time to 90% of maximum AUC/time intensity” (t90). In Fig. 4a– c, respectively, the t90 is 7.3/7.6/8.6 min in the cortex, 9.0/9.7/9.3 min in the medulla, and 9.3/10.6/10.0 min in the calyx-ureter. Agreement between the three boluses on the basis of t90 was good in all three ROIs in three out of three mice. The standard deviation (SD) in the t90 of the three boluses, when expressed as percent of the mean, was within 7.1% in all ROIs in all three mice (Table 2). These results strengthen our assumption that GdDOTP5⫺ and GdDOTA4AmP5⫺ exhibit similar kinetics of renal uptake, and also pH Imaging Using GdDOTA-4AmP5⫺ 255 FIG. 4. Representative comparison of the renal pharmacokinetics of GdDOTP5⫺ and GdDOTA-4AmP5⫺ in a female SCID mouse. Images were acquired for 1 hr each following first a bolus of GdDOTP5⫺ (0.05 mmole/Kg), then a bolus of GdDOTA-4AmP5⫺ (0.05 mmole/Kg), and then a second bolus of GdDOTP5⫺ (0.05 mmole/Kg). Shown are the data obtained following (a) the first bolus of GdDOTP5⫺, (b) the bolus of GdDOTA-4AmP5⫺, and (c) the second bolus of GdDOTP5⫺. Three ROIs within the kidney are considered: the cortex (solid), medulla (dotted), and calyx-ureter (dashed). Signal enhancement following the first GdDOTP5⫺ bolus was converted to concentration by using r1 ⫽ 3.0 mM⫺1 s⫺1. Residual enhancement from the previous bolus(es) of contrast agent was estimated by extrapolation and subtracted out prior to calculation of the relaxation rate enhancement in panel b and the AUC/time concentration of GdDOTP5⫺ in panel c. demonstrate that it is possible to reproduce the pharmacokinetics of GdDOTP5⫺ in a given animal. For the animal in Fig. 4, Fig. 5 shows an AUC/time apparent concentration map of GdDOTP5⫺ at 30 min postinjection following the first bolus (Fig. 5a), the final bolus (Fig. 5b), and the absolute difference image of the two (Fig. 5c). Excellent agreement between the two boluses can be seen for pixels falling within the kidneys, except for a small region near the proximal end of the ureter draining the right kidney, and also at the edges of the kidneys. Agreement between the two images is somewhat poorer in the extrarenal pixels, possibly due to poor signal enhancement in these regions at the dosage of contrast agent used. In Vivo pH Calculations We observed that the renal pharmacokinetics of GdDTPA2⫺ and GdDOTP5⫺ are similar (Fig. 3 and Table 1). We also observed excellent agreement between the wash-in kinetics of GdDOTP5⫺ and GdDOTA-4AmP5⫺ in the renal cortex, medulla, and calyx-ureter (Fig. 4 and Table 2 The Pharmacokinetics of GdDOTP5⫺ and GdDOTA-4AmP5⫺, and the Reproducibility of the Pharmacokinetics of GdDOTP5⫺, in SCID Mice ROI Mouse 1 Cortex Medulla Ureter Mouse 2 Cortex Medulla Ureter Mouse 3 Cortex Medulla Ureter t90 (min) GdDOTP5⫺ t90 (min) GdDOTA4AmP5⫺ t90 (min) GdDOTP5⫺ SD/ mean 7.3 9.0 9.3 7.6 9.7 10.6 8.6 9.3 10.0 0.071 0.031 0.053 8.7 9.3 9.6 8.8 10.6 11.2 8.4 9.8 10.2 0.020 0.054 0.064 8.4 9.7 10.7 7.2 8.6 10.1 7.8 8.9 9.4 0.063 0.051 0.053 Table 2). Based on these observations, and also on the similarity of molecular structure and charge between GdDOTP5⫺ and GdDOTA-4AmP5⫺ (Fig. 1), we made the assumption that the MR pharmacokinetics of GdDOTP5⫺ can be used as a surrogate for the pharmacokinetics of GdDOTA-4AmP5⫺ in a given animal. Tissue pH maps were calculated from the AUC/time at 30 min postinjection of GdDOTP5⫺ and GdDOTA-4AmP5⫺ in a given mouse, and the pH-dependent relaxivity of GdDOTA-4AmP5⫺ measured in vitro. The pharmacokinetics of the two boluses of GdDOTP5⫺ were averaged prior to calculation of pH. Figure 6b shows a representative pH image of the kidneys and nearby tissues in a control mouse. A fat-suppressed proton-density-weighted image is shown in Fig. 6a for anatomic referencing. The bulk of the pixels falling within the kidneys have a calculated pH of 7.0 –7.4, whereas the pixels corresponding to the papilla and calyx are acidic (pH 5.75–7.0). Most of the extrarenal pixels appear to be at near-neutral pH, although acidic pixels can also be seen in these regions. It should be noted that in our calculations tissue regions below ⬇pH 5.75 would be misassigned to a higher pH on the other side of the inflection in the titration in Fig. 2. From Table 1 we see that the variability between FIG. 5. Images depicting the AUC/time apparent concentration maps of GdDOTP5⫺ at 30 min postinjection for the animal in Fig. 4: (a) first bolus, and (b) last bolus. The absolute difference image 兩a ⫺ b兩 is displayed in panel c at 100% higher intensity. Pixels that were not ⱖ 5% of the signal intensity of the brightest pixel in the precontrast image, and that did not undergo at least an average of 50% enhancement in signal 5–10 min following the first bolus, were not processed and were set equal to background. 256 Raghunand et al. FIG. 6. pH image of a control female SCID mouse: (a) fat-suppressed, proton-density-weighted reference anatomic image showing a coronal view of the kidneys and nearby tissues; (b) corresponding calculated pH image. A pH of 6.0 was measured in a urine sample collected immediately following the imaging experiment. successive boluses in a given animal is of the order of 10% in the cortex and medulla, and higher in the calyx-ureter. After averaging the pharmacokinetics of the two GdDOTP5⫺ boluses, this would translate into an estimated variation of ⫺0.25 to ⫹0.20 in the calculated pH at a true pH of 7.2, for example. In three control mice, we measured a mean renal cortical pH of 7.3 ⫾ 0.13, a mean medullary pH of 7.0 ⫾ 0.29, a mean pH in the calyx-ureter ROI of 6.3 ⫾ 0.45, and a mean pH of urine collected immediately following the MR examination of 5.9 ⫾ 0.10 (Table 3). Acetazolamide, a carbonic anhydrase inhibitor, causes systemic metabolic acidosis and alkalinization of urine because the kidneys of treated mice are unable to reclaim all the filtered bicarbonate. Figure 7 shows a pH image of the kidneys and nearby tissues in an acetazolamide-treated mouse. Consistent with its expected action, acetazolamide appears to raise the pH of renal pixels by 0.4 –1.1 pH units in this animal compared to the control mouse in Fig. 6. The pH values of nonrenal tissues also appear to be more acidic than equivalent tissues in the control mouse. It must be kept in mind that the relatively low signal enhancement in nonrenal pixels means that the uncertainty in the calculated pH values in nonrenal pixels is higher than in renal pixels. Some areas of seemingly very low pH are also seen in areas of Fig. 7 with low precontrast signal intensity, such as those along the expected locations of major blood vessels or near “fatty” areas that were suppressed by fat presaturation. These artifacts likely arise FIG. 7. pH image of an acetazolamide-treated mouse. a: T1weighted, postcontrast reference anatomic image showing the kidneys and nearby tissues. The cortex, medulla, calyx, and proximal segments of both ureters are visible. b: Corresponding calculated pH image. A pH of 8.2 was measured in a urine sample collected immediately following the imaging experiment. from partial-volume effects. In two acetazolamide-treated mice the mean renal cortical pH was 7.7 ⫾ 0.18, the mean medullary pH was 8.0 ⫾ 0.13, the mean pH in the calyxureter ROI was 7.5 ⫾ 0.21, and the mean pH of urine collected immediately following the MR examination was 8.0 ⫾ 0.19 (Table 3). The increases in medullary and urine pH vis-à-vis control mice were statistically significant despite the small sample sizes. CONCLUSIONS Gene therapies show promise in the treatment of inherited and acquired renal tubular acidosis (8,9), and methods to improve the delivery and retention of corrective genes to the kidneys are being developed in animal models of the disease (29). In the near term, pH imaging applied to animal models of renal tubular acidosis will permit the noninvasive assessment of renal function, provide spatial information on the uniformity of corrective gene delivery, permit longitudinal assays of gene retention and activity on the same cohort of mice, and permit comparison of two different therapies applied to contralateral kidneys in the same animal. Potential clinical applications for pH imaging by MRI can also be envisioned for the longer term. Microenvironmental pH also impacts tumor response to chemotherapy with weakly-ionizing drugs (5), and knowledge of tumor pH distribution is important, particularly with cancers of tissues with normally acidic regions, such Table 3 Renal pH in Control and Acetazolamide-Treated SCID Mice Control Mouse 1 Mouse 2 Mouse 3 Mean ⫾ SD Acetazolamide-treated Mouse 1 Mouse 2 Mean ⫾ SD P vs. Control a pH, cortexa pH, medullaa pH, calyx-uretera pH, urine 7.11 7.27 7.44 7.3 ⫾ 0.13 6.62 7.22 7.23 7.0 ⫾ 0.29 6.88 6.34 5.78 6.3 ⫾ 0.45 5.88 6.03 5.79 5.9 ⫾ 0.10 7.47 7.82 7.7 ⫾ 0.18 0.13 7.86 8.13 8.0 ⫾ 0.13 0.04 7.25 7.66 7.5 ⫾ 0.21 0.2 7.83 8.21 8.0 ⫾ 0.19 0.001 Average pH in hand-drawn ROIs, in both kidneys and imaging slices. pH Imaging Using GdDOTA-4AmP5⫺ as renal cell carcinoma and bladder carcinoma (30). This study demonstrates the feasibility of imaging tissue pH in mice by contrast-enhanced MRI using the pH-sensitive contrast agent GdDOTA-4AmP5⫺. The pH images of acetazolamide-treated mice were consistent with the expected activity of the drug on renal tubular ion transport. Studies are underway to further characterize and minimize sources of error in the dual-contrast-agent approach reported here. ACKNOWLEDGMENTS This work was partially funded by grants from the American Cancer Society (IRG7400124-451224 to N.R.) and the NIH (R21-CA84697 to A.D.S., and R01-CA77575 and R24CA83148 to R.J.G.). REFERENCES 1. Adrogue HE, Adrogue HJ. Acid-base physiology. Respir Care 2001;46: 328 –341. 2. Adrogue HJ, Wesson DE. Acid-base. Boston: Blackwell Scientific Publications; 1994. 3. Stubbs M, Bhujwalla ZM, Tozer GM, Rodrigues LM, Maxwell RJ, Morgan R, Howe FA, Griffiths JR. An assessment of 31P MRS as a method of measuring pH in rat tumours. NMR Biomed 1992;5:351–359. 4. Raghunand N, Altbach MI, Van Sluis R, Baggett B, Taylor CW, Bhujwalla ZM, Gillies RJ. Plasmalemmal pH-gradients in drug-sensitive and drug-resistant MCF-7 human breast carcinoma xenografts measured by 31P magnetic resonance spectroscopy. Biochem Pharmacol 1999;57:309 –312. 5. Raghunand N, Mahoney B, Van Sluis R, Baggett B, Gillies RJ. Acute metabolic alkalosis enhances response of C3H mouse mammary tumors to the weak base mitoxantrone. Neoplasia 2001;3:227–235. 6. Paillard M. Na⫹/H⫹ exchanger subtypes in the renal tubule: function and regulation in physiology and disease. Exp Nephrol 1997;5:277– 284. 7. Lai LW, Erickson RP, Venta PJ, Tashian RE, Lien YH. Promoter activity of carbonic anhydrase II regulatory regions in cultured renal proximal tubular cells. Life Sci 1998;63:121–126. 8. Shayakul C, Breton S, Brown D, Alper SL. Gene therapy of inherited renal tubular disease. Am J Kidney Dis 1999;34:374 –379. 9. Lien YH, Lai LW. Gene therapy for renal diseases. Kidney Int Suppl 1997;61:S85–S88. 10. Ojugo AS, McSheehy PM, McIntyre DJ, McCoy C, Stubbs M, Leach MO, Judson IR, Griffiths JR. Measurement of the extracellular pH of solid tumours in mice by magnetic resonance spectroscopy: a comparison of exogenous 19F and 31P probes. NMR Biomed 1999;12:495–504. 11. Mason RP. Transmembrane pH gradients in vivo: measurements using fluorinated vitamin B6 derivatives. Curr Med Chem 1999;6:481– 499. 12. Van Sluis R, Bhujwalla ZM, Raghunand N, Ballesteros P, Alvarez J, Cerdan S, Galons J-P, Gillies RJ. Imaging of extracellular pH using 1H MRSI. Magn Reson Med 1999;41:743–750. 257 13. Garcia-Martin ML, Herigault G, Remy C, Farion R, Ballesteros P, Coles JA, Cerdan S, Ziegler A. Mapping extracellular pH in rat brain gliomas in vivo by 1H magnetic resonance spectroscopic imaging: comparison with maps of metabolites. Cancer Res 2001;61:6524 – 6531. 14. Vermathen P, Capizzano AA, Maudsley AA. Administration and 1H MRS detection of histidine in human brain: application to in vivo pH measurement. Magn Reson Med 2000;43:665– 675. 15. Mori S, Eleff SM, Pilatus U, Mori N, Van Zijl PCM. Sensitive detection of solvent-saturable resonances by proton NMR spectroscopy: a new approach to study pH effects. Magn Reson Med 1998;40:36 – 42. 16. Ward KM, Balaban RS. Determination of pH using water protons and chemical exchange dependent saturation transfer (CEST). Magn Reson Med 2000;44:799 – 802. 17. Zhang S, Wu K, Sherry AD. A novel pH-sensitive MRI contrast agent. Angew Chem Int Ed 1999;38:3192–3194. 18. Aime S, Botta M, Crich SG, Giovenzana G, Palmisano G, Sisti M. A macromolecular Gd(III) complex as pH-responsive relaxometric probe for MRI applications. Chem Commun 1999;16:1577–1578. 19. Mikawa M, Miwa N, Brautigam M, Akaike T, Maruyama A. Gd(3⫹)loaded polyion complex for pH depiction with magnetic resonance imaging. J Biomed Mater Res 2000;49:390 –395. 20. Beauregard DA, Parker D, Brindle KM. Relaxation-based mapping of tumour pH. In: Proceedings of the 6th Annual Meeting of ISMRM, Sydney, Australia, 1998. p 53. 21. Teicher BA, Liu S-H, Liu J-T, Holden SA, Herman TS. A carbonic anhydrase inhibitor as a potential modulator of cancer therapies. Anticancer Res 1993;13:1549 –1556. 22. Durand-Cavagna G, Owen RA, Gordon LR, Peter CP, Boussiquet-Leroux C. Urothelial hyperplasia induced by carbonic anhydrase inhibitors (CAIs) in animals and its relationship to urinary Na and pH. Fund Appl Toxicol 1992;18:137–143. 23. Chang RL, Ueki RF, Troy JL, Deen WM, Robertson CR, Brenner BM. Permselectivity of the glomerular capillary wall to macromolecules. II. Experimental studies in rats using neutral dextran. Biophys J 1975;15: 887–906. 24. Shuter B, Tofts PS, Wang S-C, Pope JM. The relaxivity of Gd-EOBDTPA and Gd-DTPA in liver and kidney of the Wistar rat. Magn Reson Imaging 1996;14:243–253. 25. Pedersen M, Morkenborg J, Jensen FT, Stodkilde-Jorgensen H, Djurhuus JC, Frokioer J. In vivo measurements of relaxivities in the rat kidney cortex. J Magn Reson Imaging 2000;12:289 –296. 26. Landis CS, Li X, Telang FW, Coderre JA, Micca PL, Rooney WD, Latour LL, Vetek G, Palyka I, Springer Jr CS. Determination of the MRI contrast agent concentration time course in vivo following bolus injection: effect of equilibrium transcytolemmal water exchange. Magn Reson Med 2000;44:563–574. 27. Rozijn TH, Van der Sanden BP, Heerschap A, Creyghton JH, Bovee WM. Determination of in vivo rat muscle Gd-DTPA relaxivity at 6.3 T. MAGMA 1999;9:65–71. 28. Aime S, Botta M, Terreno E, Anelli PL, Uggeri F. Gd(DOTP)5⫺ outersphere relaxation enhancement promoted by nitrogen bases. Magn Reson Med 1993;30:583–591. 29. Lai LW, Chan DM, Erickson RP, Hsu SJ, Lien YHH. Correction of renal tubular acidosis in carbonic anhydrase II-deficient mice with gene therapy. J Clin Invest 1998;101:1320 –1325. 30. Raghunand N, Gillies RJ. pH and drug resistance in tumors. Drug Resist Updates 2000;3:39 – 47.
© Copyright 2026 Paperzz