Scientific Report for COST Action TD 1103 (STSM) scientist: host: Johannes Scholz (PhD Candidate) Prof. Jan Henrik Ardenkjær-Larsen Technische Universität München Techincal University of Denmark Zentralinstitut für Medizintechnik Ørsteds Plads Boltzmannstr. 11 2800 Kgs. Lyngby 85748 Garching Denmark Germany Optimization of polarization of 13C Bicarbonate for in vivo ph mapping. Introduction: ph plays a major role in several biochemical processes in vivo. Hence many diseases come along with alterations in ph. As an example, many tumors have a low extracellular ph (ph=6-7) compared to the surrounding, healthy tissue due to the Warburg effect. [1] This acidification helps the tumor to degrade the environmental healthy cell matrix and leads to space in which the tumor can proliferate. Also low ph helps the tumor to overcome several therapeutic approaches due to the ph dependency of effectiveness of many anti-cancer treatments. So far, no clinical tool is established to detect and quantify ph spatially resolved. One way of overcoming this issue is the ph mapping based on hyperpolarized 13C bicarbonate, where proof on concept was shown 2008 by Gallagher et al. [2] Bicarbonate, as a major in vivo ph buffer is in chemical equilibrium with CO2, depending on the ph. The dependence is described by the Henderson-Hasselbalch equation: ph=pks+log([hco3-]/[co2]). Thus, assuming certain pks dissociation constant (in vivo=6.17 [2]), ph can be directly derived by the concentrations of bicarbonate and carbon dioxide and can be detected and mapped by magnetic resonance imaging (MRI), if the carbons are 13C labeled and hyperpolarized. This equilibrium is catalyzed in vivo by the group of enzymes called carbonic anhydrases (CA), which establish the equilibrium almost instantaneously. The technique of dynamic nuclear polarization (DNP) increases the signal of a magnetic active nucleus by a factor of more than 10^4 and the subsequent dissolution allows the detection of several biological important molecules in NMR [3]. Other diseases are also coming along with ph changes during the course of the disease, e.g. inflammations, acidosis/alkalosis and hypoxia. Like any other hyperpolarized experiment, the 13Cbicarbonate detection is limited mainly by SNR. Therefore it is necessary to increase the concentration of the hyperpolarized bicarbonate and subsequently the dissolved bicarbonate solution and optimize all of the parameters that are responsible for signal intensity, mainly polarization level and the T1 decay time of the hyperpolarized signal. Preliminary work dealt with the enrichment and up concentrating of the bicarbonate close to the solubility limits. Therefore Cesium was chosen as counter ion for synthesis due to the increased solubility of Cs-Bicarbonate (6Mol/l) compared to Na-Bicarbonate (1Mol/l). Afterwards the dissolution process was optimized at an Oxford Instruments Hyper Sense polarizer to achieve maximum dissolved bicarbonate concentration. This concentration was limited by the maximum amount that could be dissolved and a maximum of 250µl sample can be dissolved with 6ml of dissolution agent (DA). Above the amount of 250µl the
dissolution process becomes incomplete, with remaining bicarbonate in the sample cup and a resulting decrease in liquid state polarization level. The working plan was set up as week 1 should be used to increase the concentration of the dissolved bicarbonate by adjusting temperature and pressure of the dissolution process. Week 2 was planned to investigate the possibility of filtrating the OXO radical and the Cesium out of the dissolved solution and observe the influence of the filtration to the signal parameters (T1, Pol.-lvl) of the hyperpolarized Bicarbonate. Results: Sept. 2 nd to Sept. 11 th 2013: Due to the upcoming T1 question after increase of the Bicarbonate concentration, the first part was extended to the mid of the second week, as against it was planned originally. Alternative Cs-BiC preparation: Different preparation techniques have been tested to achieve best polarization level at highest Bicarbonate concentration. The already developed recipe contains enriched 13C-Cs-Bicarbonate, dissolved with same amount of a m=80%/m=20% Glycerol/D2O mixture and 20mMol/l OXO63 and 0.5mMol/l Gadolinium added. The high ratio of Glycerol leads to a sample of high viscosity at room temperature, which needs to be heated to 60 C during preparation to reduce viscosity. As a comparison, a sample with m=50%m=50% Glycerol/D2O was prepared and polarized. Advantage of this sample is, due to the higher percentage of D2O, the sample has low viscosity and no heat is needed to prepare the sample. Dissolution was performed in both cases with water + EDTA (0,1g/l). Values in brackets are from standard recipe: T1 shows comparable values of 27s (34s). However polarization level drops to 9% (19%). It can be concluded that the higher amount of Glycerol is needed to form better glassing properties and therefore the standard recipe was used for further experiments. Increase of Bicarbonate concentration in the Dissolution: Within the first week from 02.-07.09.13 the work was focused on an increase of the dissolved amount of the hyperpolarized bicarbonate. As dissolution agent usually a ph buffer based on H2O/D2O is used (e.g. TRIZ or PBS) to establish a correct ph and osmolality for subsequent in vivo injection. The dissolution agent (DA) is heated up in a bomb up to 165 C to achieve a final pressure of 10bar. Once the desired pressure is achieved, dissolution process starts. With the modified version of a Hypersense Polarizer one is able to adjust temperate and pressure to different values to achieve a higher concentration. First experiments were performed to compare polarization level and T1 times of the prepared 13C bicarbonate at the different setups. Line widths of the 11T scanner were measured with 2Hz of the CO 2 resonance peak and 6 Hz of the Bicarbonate resonance peak. Application of negative line broadening shows multiple sub peaks that indicate impurities or interactions of the Bicarbonate itself, because the broadening is only present at hyperpolarized experiments and do not show up at thermally measured experiments. Further experiments neglect this line broadening. Polarization level was measured with 19% and T1 decay was 34 seconds at 9,4T. Subsequently the increase of the dissolved bicarbonate concentration was tested. Water + EDTA (0,1g/l) was used as DA and temperature was set to 200 C. Changing the parameters, we were able to double the dissolved
amount of bicarbonate to 500µl and dissolve it successfully with 5ml of DA. This causes an increase of maximum dissolved Bicarbonate concentration to 300mMol/l. Despite the fact, that solid-state polarization was same for small volumes (3162 arb. u. @ 50µl) and big volumes (31897 arb.u. @ 500µl), liquid state polarization level dropped from 19% to 11%. An explanation for this behavior is that such big amounts of bicarbonate are dissolved in a non-optimal way. Further experiments have to be performed with pre freezing the bicarbonate solution in small droplets to increase the surface of the solution and therefore improve dissolution properties. Another observation was made for the decay in T1 relaxation, which dropped from 34s to 6,8s. Due to the fact that a T1 time of 6,8s would negate the benefit of a higher bicarbonate concentration further experiments were performed to investigate the source of the T1 time drop. Figure 1: Experimental workflow Figure 1 depicts the experimental workflow of the T1 investigation after concentration increase. The bigger boxes show key parameters for each experiment. Attached to each experiment the measured T1 time is shown. Also it is shown if the T1 was measured with hyperpolarized experiment (blue frame) or in a thermal measurement (greed frame). All experiments were carried out on a 400MHz Varian spectrometer. T1 in hyperpolarized experiments were measured with low flip angle excitation approach. Thermal T1 measurements were performed with inversion recovery pulse design, using adiabatic pulses. The pulse used was a hyperbolic secant pulse with a pulse length of 400µs. First observation was the change in T1 from low concentration and high T1 (I) to high concentration and low T1 (II). Subsequent experiments for the T1 analysis are listed clockwise for each parameter
that could influence T1. (I): Starting in the top shows the original recipe (20mMol/l OXO; 0.5mMol/l Gd), where 50µl Bicarbonate was dissolved in 5ml DA (water+0,1g/l EDTA) showing a T1 @ 9.4T of 34s. (II): Using the same Bicarbonate batch, dissolving 500µl decreases the T1 to 6,8s. (III): The increase of the dissolved sample causes a higher Gadolinium concentration and a higher OXO radical concentration. Both are substances that are essential for the polarization buildup process. To exclude Gd as the source of the T1 decrease, a 10% 13C-Cs-BiC sample was prepared at a concentration of 300mMol/l without Gd and subsequently hyperpolarized and measured. The T1 value of 8,8s shows, that also the Gd could not be the main source of the strong T1 decrease, since the expected T1 drop should be from ~30s to ~20s. (IV): Subsequent, the dissolving process was investigated as a source of T1 shortening. A sample was prepared that shows same chemical composition as a regular dissolved one, but during dissolution process no high temperature and pressure was applied. A thermal measurement showed T1 of 10s so it could be excluded as a T1 decreasing source. (V): To exclude that the used self-synthetized 13C-Cs-Bicarbonate is the source of the small T1, a thermal measurement was performed using 13Mol/l Cs-BiC, dissolved in D2O only, without Gd and OXO. The measured T1 value was 20s. (VI): To exclude the preparation procedure as a matter of influence on the chemical properties of the bicarbonate and the glycerol, where heat gun and ultrasound bath @ 60 C are used to prepare the sample, a sample with lower glycerol concentration was produced, with no heat and US-bath, which leads to a T1 of 27s. (VII): Relaxivity is also increased by the trityl radical OXO63 which concentration is 0,17mMol in the 50 µl sample and 1.7mMol in the 500µl sample. It should increase the relaxivity of the 13C atom by 0.01/mM/s. This should decrease the T1 from ~30s to ~20s and could not explain the huge decrease in T1. To check the influence of the radical in high field, thermal T1 measurements were performed at different OXO concentrations ranging from 0 to 2 mmol/l. T1 dropped from 30s to 15s which is still more than 2 times higher that the observed T1 of 6.8s and can be therefore not the only cause of the T1 drop. To take into account that relaxivity increases higher at low field strengths, field cycling experiments were performed to evaluate also the low field influence of OXO to the T1. Same samples of the high field measurements lead to doubtable values that do not show a consistent decrease in T1 for increasing OXO radical concentrations. Further investigations need to be performed on that side. In the end it could not completely figured out what is the reason of the unexpected high T1 time decrease. One assumption could be interplay of the different parameters. Additionally chemical shift anisotropy could be an explanation due to increased viscosity of the higher concentrated sample. Sept. 12 th to Sept. 13 th 2013: Filtration: Main source of T1 decay at such high concentrations is the paramagnetic influence of the OXO radical. To avoid its effect on T1 it would be preferable to remove the radical as soon as possible from the dissolved liquid. Its removal is also needed for further in vivo application and/or later clinical studies. Also the cesium, needed as a counter Ion for the synthesis and up concentrating, should be removed and exchanged with less low toxic counter ions to form sodium bicarbonate with a low LD50 of 4220mg/kg(oral) [4]. The hyperpolarized liquid was dissolved and cleared by anionrespectively cation exchange columns and polarization level and T1 were measured. Amount of resin was estimated by the amount of compound to be filtered. Cesium:
Cesium concentration was detected with a tunable probe at 9,4T at 133Cs resonance frequency. To remove the Cesium a cation exchange column (DOWEX 50Wx8) with a mesh size of 200-400 was used in hydrogen form. Cesium concentration went down from 15mMol/l below the detection limit of the scanner and is therefore assumed as successfully cleared. Polarization level (22%) and T1 (37s) were not affected by the Cesium-Sodium exchange. OXO: For OXO radical clearance test an anion exchange column containing DOWEX50Wx8 200-400 mesh size was used. To decrease the affinity to the bicarbonate the resin was treated over night with 1M Na2SO4. After dissolution the liquid was pressed through the resin, T1 and Polarization level were measured and the concentration of the bicarbonate was estimated. OXO itself has a dark green color and even low concentrations of the radical dye liquids intensively green. Hence the color of the liquid gives evidence about the radical concentrations of the corresponding solution. After filtration with the exchange column the liquid moved from green to completely colorless by visual inspection. Also T1 increased to 40s and polarization level was 22.4%, so it can be concluded that the entire radicals were successfully removed. Filtration has no negative effect on polarization level and T1. Despite the Na2SO4 treatment, Bicarbonate concentration dropped from 15mMol/l to 0,06mMol/l after filtration, so the affinity of the resin is still high for bicarbonate. Further investigations need to be carried out with C18 Anion Exchange resin that should have no affinity to the bicarbonate and high affinity to the OXO radical. Conclusion and discussion: During the two weeks of research stay the maximum concentration of the dissolved bicarbonate could be more than doubled to 300 mmol/l. Nevertheless the huge decrease in T1 partly undoes the benefit of the larger concentration. All possible sources of the T1 decrease were investigated and quantified. As a result each single parameter itself does not have enough influence to cause the fast decay alone. Further investigations need to be performed to look for interplay of the different parameters. Cesium filtration was successful and shows no negative effects on the hyperpolarized signal and could therefore be used for further preclinical and clinical applications. Filtration of the OXO radical was also successful with huge benefits in T1 and subsequent signal conservation. Nevertheless the high affinity of the resin for the bicarbonate needs be overcome by use of a different resin or reducing affinity of the resin for the bicarbonate. Acknowledgement/References: [1] Otto Warburg, Science 123, no. 3191 (February 24, 1956): 309-314. [2] Ferdia A. Gallagher et al., Nature 453, no. 7197 (June 12, 2008): 940-943. [3] Jan H. Ardenkjær-Larsen et al., Proceedings of the National Academy of Sciences 100, no. 18 (2003): 10158-10163. [4] http://chemlabs.uoregon.edu/safety/toxicity.html