Technical Note: A Deuterated ¹³C-Urea Reference for Clinical Multiparametric Prostate Cancer Studies Including Hyperpolarized Pyruvate

Abstract

Purpose:

Metabolic MRI using hyperpolarized [1-¹³C]-pyruvate offers unprecedented new insight into disease and response to therapy. ¹³C-enriched reference standards are required to enable fast and accurate calibration for ¹³C studies, but care must be taken to ensure that the reference is compatible with both ¹³C and ¹H acquisitions. The goal of this study was to optimize the composition of a ¹³C-urea reference for a dual-tuned ¹³C/¹H endorectal coil and minimize imaging artifacts in metabolic and multiparametric MRI studies involving hyperpolarized [1-¹³C]-pyruvate.

Methods:

Due to a high amount of Gd doping for the purpose of reducing the spin-lattice relaxation time (T1) of urea, the ¹H signal produced by a reference of ¹³C-urea in normal water was rapidly relaxed, resulting in severe artifacts in heavily T1-weighted images. Hyperintense ringing artifacts in ¹H images were mitigated by reducing the ¹H concentration in a ¹³C-urea reference via deuteration and lyophilization. Several references were fabricated and their SNR was compared using ¹H and ¹³C imaging sequences on a 3T MRI scanner. Finally, ¹H prostate phantom imaging was conducted to compare image quality and ¹H signal intensity of normal and deuterated urea references.

Results:

The deuterated ¹³C-urea reference provides strong ¹³C signal for calibration and an attenuated ¹H signal that does not interfere with heavily T1-weighted scans. Deuteration and lyophilization were fundamental to the reduction of ¹H signal and hyperintense ringing artifacts. There was a 25-fold reduction in signal intensity when comparing the non-deuterated reference to the deuterated reference, while the ¹³C signal was unaffected.

Conclusion:

A deuterated reference reduced hyperintense ringing artifacts in ¹H images by reducing the ¹H signal produced from the ¹³C-urea in the reference. The deuterated reference can be used to improve anatomical image quality in future clinical ¹H and hyperpolarized [1-¹³C]-pyruvate MRI prostate imaging studies.

Introduction

The ability to manage and predict outcome for patients with prostate cancer is a critically important unmet clinical need. Currently, a crucial goal for prostate cancer imaging is accurate disease characterization, which can ultimately be achieved through a combination of multi-modality information gathered by minimally invasive imaging techniques.¹ The metabolic properties of localized prostate cancer are of great interest due to the high rate of aerobic glycolysis undergone by prostate tumors.^2,3 Hyperpolarized (HP) ¹³C magnetic resonance imaging (MRI) is an emerging metabolic imaging method that can be used to quantify the conversion of HP pyruvate into lactate as an indicator for glycolytic flux in vivo.⁴ HP ¹³C MRI can be paired with dynamic contrast-enhanced (DCE-) MRI to inform on vascular function and enhance quantitative analysis of ¹³C signal evolution.⁴ HP ¹³C MRI and DCE-MRI have the ability to provide attractive multiparametric imaging for patients with prostate cancer. In addition to standard metabolic information provided by HP [1-¹³C]-pyruvate for prostate cancer studies,^5,6 anatomical and DCE-MRI can be performed during the same examination, without moving the patient.

Hyperpolarization of ¹³C labeled agents can result in a greater than 100,000-fold gain in signal intensity compared to thermal equilibrium;⁷ it also enables real-time, in vivo observation of biochemical processes, thus allowing investigators the ability to probe specific enzymatic pathways. ¹³C-labeled pyruvate and its downstream metabolites have been shown to be of interest regarding the characterization of tumor metabolism.^8–10 Prior to injection of HP [1-¹³C]-pyruvate, it is necessary to calibrate to the ¹³C center frequency and transmit power for imaging. This can be accomplished by using a ¹³C-labeled urea reference placed inside the coil.¹¹ The most common way to create a ¹³C-urea reference is by dissolving ¹³C-urea in H₂O. In this way, the reference can provide ¹³C and ¹H signals, which can be detected using ¹³C/¹H dual-tuned coils.

Currently, dual-tuned ¹³C/¹H endorectal coils (ERC) are used for clinical assessment of HP [1-¹³C]-pyruvate in the setting of prostate cancer,^5,12–14 as the use of dual-tuned ¹³C/¹H coils improves acquisition efficiency, quantitative analysis, and allows for the collection of ¹H images and ¹³C spectral data.¹⁵ A ¹³C-urea reference standard can be integrated into ¹³C/¹H ERC. However, due to the close proximity to coil elements and high amount of Gd doping for the purpose of reducing the spin-lattice relaxation time (T1) of ¹³C-urea, the ¹H signal produced by the reference is rapidly relaxed and can be much greater than that of nearby tissues. As a result, the large signal produced by the reference can lead to imaging artifacts that greatly reduce ¹H image quality, thus limiting the diagnostic utility of ¹H anatomical scans. Therefore, the purpose of this research was to develop a dual-tuned ¹³C-urea reference containing optimized reagents. Optimal ¹³C-urea reference fabrication was ultimately achieved by replacing H₂O with D₂O and reducing the ¹H signal of urea by deuteration and lyophilization of ¹³C-urea in D₂O.^16,17

Materials and Methods

¹³C-Urea Reference Fabrication

Five candidate ¹³C-urea references were created for comparison to determine the optimal mixture that would reduce ¹H signal intensity. Three references were fabricated by dissolving 244 mg of non-lyophilized, once lyophilized, or twice lyophilized ¹³C-urea (Cambridge Isotope Laboratories, Andover, MA, USA, Item Number: CLM-311-PK) to create 4M ¹³C-urea solutions in 1.00 mL D₂O (99.9 atom % D, Sigma-Aldrich, Milwaukee, WI, USA, SKU: 151882). A fourth reference was fabricated by dissolving 244 mg non-lyophilized ¹³C-urea in 1.00 mL distilled H₂O. The fifth reference was comprised by dissolving 488 mg ¹³C-urea to create an 8M ¹³C-urea reference in a mixture of 0.10 mL glycerol and 0.90 mL H₂O. Following lyophilization of ¹³C-urea in D₂O, all five references were also mixed with 0.65 mg of NaN₃ (Fisher Scientific, Fair Lawn, NJ, USA, Catalog Number: S227I-25) to obtain a 10 mM NaN₃ concentration and 3 μL 1 mmol/mL Gadovist contrast agent (Bayer HealthCare Pharmaceuticals Inc., Whippany, NJ, USA). It is important to note that the addition of Gadovist contributed to additional ¹H signal in the references. After vortexing, each sample was added to a 1 mL syringe, which was sealed with hot glue.

Lyophilization was performed as follows: two aliquots of 244 mg ¹³C-urea were dissolved in 1.00 mL D₂O each. After approximately one hour in room temperature, both samples were frozen in liquid nitrogen and lyophilized overnight with a freeze dryer (Labconco FreeZone 4.5 Liter −105°C, Labconco Corporation, Kansas City, MO, USA, Catalog Number: 7382020). This process was repeated for one sample after redissolving in 1.00 mL D₂O, resulting in a twice lyophilized sample of ¹³C-urea.

Magnetic Resonance Imaging Acquisition

All images were acquired with a GE Discovery 3T scanner (MR750, GE Healthcare, Waukesha, WI, USA). To begin, the non-lyophilized urea in H₂O, non-lyophilized urea in D₂O, once lyophilized, and twice lyophilized deuterated urea references were scanned together in order to compare their relative signal intensity. It was quickly determined that the non-lyophilized urea in H₂O and in D₂O references severely dominated the dynamic range of the image. Those references were removed and the scan was repeated for the lyophilized references in D₂O. The references were scanned using a ¹H 3D fast spoiled gradient echo sequence (FSPGR), similar to our clinical DCE-MRI sequence, inside a ¹³C/¹H volume resonator with a 35-mm inner diameter (RAPID Biomedical GmbH, Rimpar, Germany). The ¹H 3D FSPGR images were acquired with a 3.87 ms repetition time, 1.41 ms echo time, 22 cm × 22 cm field of view, 3.6 mm slice thickness, and a 20-degree excitation angle.

A ¹³C gradient echo-planar imaging (EPI) sequence¹⁸ was performed for the reference configuration containing all four references to ensure that the references provided similar ¹³C signal, regardless of if they produced ¹H imaging artifacts. ¹³C EPI was acquired with a 1000 ms repetition time, 25 ms echo time, 6 cm × 6 cm field of view, a single NEX, 40 mm slice thickness, and a 90-degree excitation angle centered on the resonance frequency for ¹³C-urea, as determined from a non-selective pulse-acquire spectrum.

Next, ¹H 3D FSPGR of a prostate phantom (Figure 1) was acquired using the dual tuned ¹³C/¹H ERC (RAPID Biomedical GmbH, Rimpar, Germany) containing the non-deuterated reference (488 mg ¹³C urea dissolved in a mixture of 0.10 mL glycerol and 0.90 mL H₂O). This urea reference was then replaced with the twice lyophilized deuterated reference and the ¹H 3D FSPGR scan was repeated. The ¹H 3D FSPGR of a prostate phantom for both the non-deuterated urea reference and twice lyophilized deuterated urea reference were acquired with a 3.07 ms repetition time, 1.34 ms echo time, 22 cm × 22 cm field of view, 3.6 mm slice thickness, left-right frequency encoding direction, and a 20-degree excitation angle. To permit quantitative comparison of the relative signal from these references, the mean signal from a region-of-interest (ROI) encompassing the reference was normalized by the root-mean-square noise amplitude measured in an artifact-free background region of the image.

Figure 1.

Endorectal coil, ¹³C-urea reference, and prostate phantom setup. (a) The ¹³C-urea sample, located inside of a 1 mL syringe, can easily be integrated into the ERC. (b) The head of the ERC contains several regions that house coil elements, as well as a bored out region in the center of the coil head where the ¹³C-urea reference resides. (c) A prostate phantom setup was used to compare the signal intensity of the non-deuterated and deuterated ¹³C-urea references.

¹³C T1 Measurements

¹³C T1 measurements were collected using the GE Discovery 3T scanner inside the ¹³C/¹H volume resonator. Four reference samples were measured individually: non-lyophilized urea in H₂O without Gd, non-lyophilized urea in H₂O with Gd, twice lyophilized urea in D₂O without Gd, and twice lyophilized urea in D₂O with Gd. A non-selective spectroscopy method was used to acquire variable flip angle measurements for calculating the T1 for ¹³C-urea in these mixtures.

Results

¹H 3D FSPGR MRI acquired of the ERC embedded with the non-deuterated urea reference inside a prostate phantom (Figure 2) showed hyperintense ringing artifacts in both the phase and frequency encoding directions. This reference yielded a normalized signal of 2.77 × 10³.

Figure 2.

¹H 3D FSPGR MRI acquired with ERC containing the non-deuterated ¹³C-urea reference inside a prostate phantom. There is a clear ¹H hyperintense ringing artifact produced by relaxed ¹³C-urea in H₂O. The red ROI is the area used to calculate the normalized signal intensity of the reference.

¹H 3D FSPGR MRI of all references (Figure 3a), acquired using the ¹³C/¹H volume resonator, demonstrated the hyperintensity of the non-lyophilized urea in distilled H₂O (labeled as reference four in Figure 3a) as it dominated the dynamic range compared to the other three urea in D₂O references, especially the two lyophilized references (labeled as reference one and two in Figure 3a). ¹H 3D FSPGR MRI of the non-lyophilized, once lyophilized, and twice lyophilized urea in D₂O references (Figure 3c) demonstrated that even after removing the non-lyophilized urea in distilled H₂O reference, the dynamic range was dominated by the non-lyophilized urea in D₂O reference (labeled as reference three in Figure 3c). ¹H 3D FSPGR MRI of the once lyophilized and twice lyophilized references (labeled as reference one and two in Figure 3d) was the first image where more than one reference was clearly visible (Figure 3d). ¹³C T1 measurements showed that deuteration of ¹³C-urea increased the ¹³C T1 from 37.7s to approximately 96.7s, but this effect was mitigated by the addition of Gd which reduced the ¹³C T1 to approximately 0.474s (compared to 0.545s for non-lyophilized urea in H₂O). ¹³C EPI of all references (Figure 3b) showed that the ¹³C signals detected from the references were comparable. The normalized ¹H signal intensity of the non-lyophilized urea in H₂O, non-lyophilized urea in D₂O, once lyophilized, and twice lyophilized urea references in D₂O were 2.80 × 10⁴, 2.84 × 10³, 5.65 × 10², and 3.62 × 10², respectively.

Figure 3.

¹H 3D FSPGR MRI and ¹³C EPI of the samples using a ¹³C/¹H volume resonator. (a) ¹H MRI was conducted for the non-lyophilized ¹³C-urea in distilled H₂O (four), non-lyophilized ¹³C-urea in D₂O (three), twice lyophilized ¹³C-urea in D₂O (two), and once lyophilized ¹³C-urea in D₂O (one) reference group. (b) ¹³C EPI was conducted for the non-lyophilized ¹³C-urea in distilled H₂O, non-lyophilized ¹³C-urea in D₂O, twice lyophilized ¹³C-urea in D₂O, and once lyophilized ¹³C-urea in D₂O reference group in the same position as (a). (c) ¹H MRI was conducted for the non-lyophilized ¹³C-urea in D₂O, twice lyophilized ¹³C-urea in D₂O, and once lyophilized ¹³C-urea in D₂O reference group. (d) ¹H MRI was conducted for the twice lyophilized and once lyophilized ¹³C-urea in D₂O references.

¹H 3D FSPGR MRI acquired of the ERC embedded with the twice lyophilized reference (Figure 4) showed reduction of the hyperintense ringing artifacts seen in Figure 2. The normalized signal intensity of the twice lyophilized reference was 1.13 × 10². Therefore, the signal intensity of the non-deuterated urea reference compared to the twice lyophilized deuterated reference showed a reduction by a factor of approximately 25.

Figure 4.

¹H 3D FSPGR MRI acquired with ERC containing the deuterated urea reference inside a prostate phantom. The hyperintense ringing artifact has been reduced as a result of the twice lyophilized deuterated reference replacing the non-deuterated urea reference. The red ROI is the area used to calculate the normalized signal intensity of the reference.

Discussion

Replacement of H₂O with D₂O in the reference was a fundamental first step toward the reduction of hyperintense ringing artifacts produced during ¹H imaging. However, this action alone did not solve the problem. The exchange of the ¹H located on the urea molecule with the surrounding D₂O creates H₂O in the surrounding solvent. Therefore, urea in D₂O alone is insufficient to reduce the hyperintense ringing artifacts because there is still a substantial ¹H signal that arises from the urea and surrounding D₂O solvent. [D₄, ¹³C]-urea can be created in the laboratory and is a much better alternative when dissolved in D₂O because it does not increase ¹H signal in the reference. In this work, we found that two cycles of exchange in D₂O were sufficient to attenuate the heavily doped ¹H signal in the reference to be consistent with signal observed in a tissue-mimicking phantom.

It is important to note that two rounds of lyophilization, in order to achieve optimal ¹H signal reduction, is not required. This experiment was carried out by performing two rounds of deuteration and lyophilization of 244 mg ¹³C-urea in 1 mL of D₂O to demonstrate the progressive reduction in ¹H signal. However, the same fractional enrichment of amine hydrogen atoms could be achieved by a single lyophilization of 244 mg ¹³C-urea in 8.75 mL of 99.9% D₂O.

The twice lyophilized deuterated urea reference successfully reduced the ¹H imaging artifacts produced by the highly Gd doped, undeuterated ¹³C-urea reference, to below the noise floor of our FSPGR acquisition. The hyperintense ringing artifacts observed from the undeuterated reference limit the diagnostic quality of the ¹H images and DCE-MRI scans acquired in conjunction with HP ¹³C MRI. The twice lyophilized reference was able to successfully reduce the mean signal intensity, which led to the reduction of hyperintense ringing artifacts. Furthermore, the ¹³C-urea signal from the twice lyophilized reference was unaffected, allowing it to continue to serve as a calibration reference for future ¹³C studies. The ¹³C T1 of non-lyophilized urea in H₂O with Gd and twice lyophilized urea in D₂O with Gd were comparable, and short enough that either can successfully be used as a convenient reference for ¹³C calibrations. The deuterium-enriched urea reference is a fundamental ¹³C calibration tool which will not produce imaging artifacts during ¹H MRI and thus will support the acquisition of anatomical and functional data to compliment HP ¹³C imaging.

The twice lyophilized urea reference is not only useful for prostate cancer imaging and implementation in ERC but can also be applied to HP ¹³C imaging for other anatomical sites and coil configurations. Any laboratory that uses dual-tuned ¹³C/¹H coils for anatomic, functional, and HP ¹³C MRI could integrate a similar calibration reference to reduce imaging artifacts that stem from excessive ¹H signal in their reference. Uncompromised ¹H imaging will allow for standard of care imaging alongside HP ¹³C imaging studies. Furthermore, multiparametric ¹H studies can improve the quantification of ¹³C.⁴

Conclusion

We designed a reference standard for calibration of ¹³C scans that does not compromise anatomical or functional ¹H MRI. The deuterated urea references showed a reduction in signal intensity when compared to a non-deuterated reference embedded in the ERC, and reduced ¹H ringing artifacts in heavily T1-weighted images. The deuterium-enriched ¹³C-urea reference can be used to improve anatomical image quality in future clinical ¹H and HP [1-¹³C]-pyruvate MRI prostate cancer imaging studies.