Abstract
Common medical imaging techniques usually employ contrast agents that are chemically labeled, e.g. with radioisotopes in the case of PET, iodine in the case of CT or paramagnetic metals in the case of MRI to visualize the heterogeneity of the tumor microenvironment. Recently, it was shown that natural unlabeled D-glucose can be used as a nontoxic biodegradable contrast agent in Chemical Exchange sensitive Spin-Lock (CESL) magnetic resonance imaging (MRI) to detect the glucose uptake and potentially the metabolism of tumors. As an important step to fulfill the clinical needs for practicability, reproducibility and imaging speed we present here a robust and quantitative T1ρ-weighted technique for dynamic glucose enhanced MRI (DGE-MRI) with a temporal resolution of less than 7 seconds. Applied to a brain tumor patient, the new technique provided a distinct DGE contrast between tumor and healthy brain tissue and showed the detailed dynamics of the glucose enhancement after intravenous injection. Development of this fast and quantitative DGE-MRI technique allows for a more detailed analysis of DGE correlations in the future and potentially enables non-invasive diagnosis, staging and monitoring of tumor response to therapy.
Similar content being viewed by others
Explore related subjects
Discover the latest articles, news and stories from top researchers in related subjects.Introduction
In diagnostic imaging it is often beneficial to enhance the contrast in tissue or to make a contrast more specific to a certain physiology or pathology. This is usually achieved by chemical labeling of specific agents, for example by labeling metabolites with radioisotopes in the case of PET, making use of iodinated compounds in the case of CT or using chelated paramagnetic metals in the case of MRI. As a paradigm shift, it was shown that natural unlabeled D-glucose could serve as a biodegradable contrast agent for the detection of cancer by employing chemical exchange saturation transfer (CEST) or chemical exchange sensitive spin-lock (CESL) magnetic resonance imaging (MRI). Labeling in the case of CEST MRI works non-invasively by selective radiofrequency (rf) irradiation: e.g. hydroxyl protons of glucose are labeled by means of rf irradiation that matches their chemical shift and their proton exchange regime. This labeling is transferred to water protons by chemical exchange and can be detected via MRI. The feasibility to track the uptake of glucose in animals was proven employing both techniques, CEST1,2,3,4,5 and CESL6,7,8. First results in human tumor patients were recently published by Xu et al. and Wang et al. by means of CEST9,10 and by Schuenke et al. employing an adiabatically prepared CESL technique11.
However, the history of many MRI contrasts showed that the translation of new contrasts into clinical routine requires a fast and robust technique and an evaluation process, which is as simple as possible. To fulfill those clinical needs, we show herein that CESL-based dynamic glucose enhanced MRI (DGE-MRI) can be accelerated essentially and made robust against field inhomogeneities by means of adiabatically prepared T1ρ-weighted (T1ρ-w) imaging. Further, we introduce a simple but appropriate evaluation method that provides a quantitative T1ρ-w DGE contrast. In simulations and in vitro experiments, we demonstrate that the proposed contrast depends linearly on glucose concentration changes and is independent of tissue-specific relaxation parameters. After implementation and optimization at a 7T MRI scanner, this technique was employed in a glucose-injection experiment with seven-second temporal resolution. First results of T1ρ-w DGE-MRI in a patient with glioblastoma are presented revealing a substantial DGE contrast between tumor and healthy tissue.
As the origin of the DGE contrast is still under discussion1,2,3,4,5,6,7,8,9,10,11, the presented technique does not only form a simple and robust diagnostic tool for studying the DGE contrast in clinical studies, but with its high temporal resolution also serves as a research tool. Thus, it might help solving the question to what extent the occurring contrast originates from intra- or extracellular glucose level changes, from pH changes, or from other glucose related metabolites.
Results
T1ρ-weighted DGE-MRI in vivo
The accelerated and quantitative T1ρ-w DGE-MRI protocol optimized with respect to contrast and scanning time was applied with temporal resolution of less than 7 seconds in the study of a patient with a brain tumor. The tumor (glioblastoma, WHO grade IV) located in the left frontal lobe can be identified in the T2-w image acquired at 7T (Fig. 1a) and in the co-registered gadolinium contrast-enhanced T1-weighted (GdCE-T1w) image obtained at 3T (Fig. 1b). We define the quantitative T1ρ-weighted dynamic glucose enhancement (DGEρ) by the relative signal difference (Eq. 4 in Methods) at each time point
for evaluation of series of T1ρ-w in vivo images. DGEρ depends linearly on glucose concentration changes and is independent of tissue-specific relaxation parameters as demonstrated in simulations and in vitro experiments (see below). DGEρ was calculated for each time point in every voxel employing the average of 18 T1ρ-w images acquired before start of the glucose injection as reference (Sref). First of all, the DGEρ images obtained after glucose injection (cf. Fig. 1c, t = 588 s) clearly delineate the tumor region consistent with the GdCE-T1w image (Fig. 1b). We further evaluated DGEρ as a function of time in three regions of interest (ROIs), namely a tumor-ROI (ROI #1) selected on the DGEρ image shown in Fig. 1c, a second tumor-ROI (ROI #2) selected on the GdCE-T1w image (Fig. 1b), and a ROI in normal appearing white matter (ROI #3). The ROI-specific DGEρ curves are shown in Fig. 1d. The ROIs are marked in the GdCE-T1w and DGEρ image shown in the top left corner. Before the start of glucose injection (at t = 0 s) DGEρ of all three ROIs fluctuated around 0% and consequently no tumor contrast was visible in the corresponding DGEρ image shown in Fig. 1e. After the start of injection, all curves slightly increased and a faint contrast became apparent in the tumor area as well as in the paraventricular area about 1 min after the end of the injection phase (Fig. 1f). For t ≥ 200 s the curves of both tumor ROIs were outside the error of normal appearing white matter (ROI #3), which showed only a minor increase in the DGEρ curve over the entire time course. Accordingly, the contrast in the DGEρ images increased, revealing another slightly enhancing region (black arrow) at the bottom of the tumor area (Fig. 1g), which remained visible in the DGEρ images obtained afterwards (Fig. 1h and i). The highest contrast was observed at about 10 min after start of the injection (Fig. 1h), where the T1ρ-w dynamic glucose contrast in ROI #1 was more than twice as high compared with that in ROI #2 and about 8 times higher than in normal appearing white matter. The subsequent signal drop in the DGEρ curve (red arrow in Fig. 1d) was most likely due to patient motion, which was identified by a displacement of the brain position in the time-resolved T1ρ-w images. Interestingly, the DGEρ images did not show any contrast in blood vessels.
Bloch-McConnell simulations
To investigate the contrast obtained with T1ρ-w MRI we simulated T1ρ relaxation curves by means of a Bloch-McConnell simulation tool. Figure 2a shows the relaxation curves for glucose concentrations of 5 mM and 20 mM and transversal relaxations rates R2 = 15 s−1, 20 s−1 and 25 s−1. Figure 2b displays the difference ΔS between the simulations for 5 mM and 20 mM for the three R2 (solid lines) together with analytical approximations (dashed lines) obtained with equation (3) (see in Methods below). The approximation agrees well; especially the maxima appear at the same position proving its validity. As predicted by equation (3) the curves differ for varying R2 making ΔS an inappropriate measure for changes of glucose concentration in vivo. However, the approximation suggests that this dependency on R2 can be eliminated by dividing equation (3) by the reference signal . This yields the relative signal change ΔSrel given by equation (4), that also defines DGEρ(t) (Eq. 1) at a specific time point t. ΔSrel, and thus DGEρ(t) depends only on TSL and the variation ΔRex of the exchange-dependent relaxation rate Rex. In Fig. 2c ΔSrel is plotted as a function of the glucose concentration change ΔcGlc for R2 = 15 s−1 (solid blue line) and R2 = 25 s−1 (green diamonds) for one specific spin-lock time of 50 milliseconds. The plot shows that the contrast does not depend on R2. The analytical approximation (Eq. 4; dashed red line) agrees well again. Thus, DGEρ(t) defined by ΔSrel represents a quantitative contrast, which depends linearly on ΔRex and hence on changes of the glucose concentration (ΔcGlc) for a given TSL. To determine the optimum TSL (TSLopt) one has to consider not only the signal-to-noise ratio (SNR), but also the contrast-to-noise ratio (CNR). Assuming a constant SNR, the maximum CNR is given by the position of the maxima of ΔS. Equation (3) allows to determine this point analytically yielding TSLopt =T1ρ. For the simulated relaxation rates, which represent the range we observed in human brain tissue at 7T using the adiabatically prepared spin-lock approach11, the CNR for TSL = 50 ms (red vertical line in Fig. 2b) is close to the optimal value for all relaxation rates that we considered.
In vitro experiments
To confirm the results of our simulations we performed measurements of aqueous solutions with different glucose concentrations and different R1 and R2. The relaxation rates were adapted using gadoteric acid and agar for one set of solutions and MnCl2 for a second set. In the following, the different sets are called agar phantoms and MnCl2 phantoms, respectively. The measured T1ρ relaxation curves for glucose concentrations of 20 mM (solid lines) and 40 mM (dashed lines) are plotted in Fig. 3a for the agar and the MnCl2 phantoms. The curves were normalized to the first value and all data points represent the mean and standard deviation of three independent measurements. Figure 3b shows the signal differences ΔS between the two particular relaxation curves from Fig. 3a. The curves display the dependence on the absolute relaxation rates expected from the simulations (Fig. 2b) and equation (3). Figure 3c shows the relative signal difference ΔSrel for a constant spin-lock time of 100 milliseconds as a function of the glucose concentration change. The curves for both, the agar and the MnCl2 phantoms, agree within the errors (shown only for the MnCl2 phantom measurements for the sake of clarity). The consistency of both curves proves the independence of the relative signal difference ΔSrel (or rather DGEρ for series of in vivo images) on absolute relaxations rates.
Discussion
In this study, we showed that T1p-based DGE-MRI can be accelerated essentially by employing T1ρ-w imaging. The introduced contrast called T1ρ-w dynamic glucose enhancement (DGEρ, Eq. 1) was shown to be independent of relaxation parameters of tissue and direct proportional to changes of the glucose concentration thus enabling fast and quantitative DGE-MRI in a glioblastoma patient with a temporal resolution of less than 7 seconds.
So far glucose enhanced MRI in humans has been performed in brain tumor patients at 7T by means of CEST9 and T1ρ mapping11 and in head and neck tumor patients at 3T by means of CEST10. In all studies, an increased glucose uptake was reported after intravenous injection of natural D-glucose. However, the studies substantially differed in the temporal resolution, varying between 5 seconds in the case of CEST-based dynamic glucose enhanced MRI applied by Xu et al.9 and about 5 minutes in the study of Wang et al.10. The temporal resolution of the T1p-weighted approach proposed in this study is below seven seconds and thus in the same order as for CEST-based DGE-MRI. As the spin-lock preparation time of 50 ms is much shorter compared to CEST saturation, which normally requires seconds, the temporal resolution can be increased to about 3 s if SNR is sufficient.
High temporal resolution is mandatory to detect variations on small time scales like changes in the blood glucose level (BGL) after a bolus glucose injection. Robust tracking of the BGL could potentially enable pharmacokinetic modelling based on compartment models as for example employed in gadolinium-based dynamic contrast enhanced MRI (DCE-MRI)12. Another benefit of a high temporal resolution is the opportunity to increase the effective SNR and CNR by averaging of several consecutive measurements. This could be relevant for glucose enhanced MRI when a lower temporal resolution is sufficient, e.g. when the bolus injection is replaced by a continuous glucose infusion, but also for native T1ρ-based imaging without glucose enhancement. Consequently the presented adiabatically prepared T1ρ-w imaging technique with the proposed normalization might also improve cartilage imaging, where T1ρ mapping is a common technique to detect the loss of proteoglycan in the early stages of osteoarthritis13,14,15,16.
As shown previously, an adiabatically prepared spin-lock approach combined with a non-adiabatic MRI readout, as used in our study, works within specific absorption rate (SAR) restrictions and technical limitations of ultrahigh field whole-body scanners11. This leads to a homogenous T1ρ contrast over the entire brain despite B1 inhomogeneities and consequently to negligible contributions from B1 dispersion to the DGE contrast11. We want to point out that for the in vivo T1ρ-w DGE-MRI measurement, SAR was around 50% of the allowed value and hence relatively low for using adiabatic pulses. This can be understood since only two adiabatic half-passage pulses are used per 7 s. Consequently, a reduction of the recovery time and thus an increase of the temporal resolution is also in accordance with SAR restrictions. Furthermore, the proposed T1ρ-w DGE-MRI inherits all benefits of the adiabatically prepared spin-lock approach. This includes the higher sensitivity to the intermediate and fast exchange regime relevant for glucose and the enhanced robustness against B0 inhomogeneities compared to CEST6,17, but also the fact that changes in DGEρ due to inhomogeneities in the B1 field are negligible compared to changes induced by variations of the glucose concentration11. Especially the robustness against field inhomogeneities qualifies the presented approach for application at whole-body ultra-high field scanners. These are of great interest for chemical exchange sensitive experiments due to the increasing exchange-weighting with higher field strength18,19. Further, the robustness against field inhomogeneities makes the application of correction methods dispensable and thus simplifies the post-processing.
As predicted by our analytical approximation (Eq. 3) we could show that the dependency of the signal difference (ΔS) on absolute relaxation rates can be eliminated by an appropriate normalization yielding the T1ρ-weighted dynamic glucose enhancement (DGEρ), which depends linearly on the glucose concentration and is independent of relaxation parameters of the tissue. These properties could be verified with simulations (Fig. 2c) and in vitro measurements (Fig. 3c). We want to point out that the intrinsic robustness of the adiabatic spin-lock against field inhomogeneities in combination with the introduced normalization yield a quantitative contrast, which can be compared between different measurements and subjects. CEST-based DGE-MRI techniques, on the other hand, can be prone to influences of inhomogeneities and absolute relaxation rates20. Although faster T2 relaxation due to exchange21 can lead to a signal enhancement in CEST-based DGE-MRI, being independent of T1 and T2 relaxation has the benefit of the above mentioned quantitative contrast and additionally some practical benefits: with DGEρ it is possible to perform a DGE measurement after gadolinium injection which is practical in clinical routine. Beyond that, it is also thinkable to perform DGE and DCE with the same injection bolus at the same time, which would speed up the acquisition and provides a reference for pharmacokinetic investigations.
However, also in the case of CEST the influences of absolute relaxation rates can be handled by employing relaxation compensation techniques. Figure 4 shows simulated T2- and T1-dependencies of the CEST-based DGE approach based on the paper of Xu et al.5. Our simulations reveal that by using a T1 map and the AREX22 metric, also quantitative CEST-based DGE-MRI can be realized.
DGEρ in brain tumor patient
Applying the proposed quantitative T1ρ-w DGE-MRI approach with seven-second temporal resolution evaluated using DGEρ in a glioblastoma patient we observed an increasing DGE contrast in the tumor area after the intravenous glucose bolus injection. This finding is in agreement with the outcome of our previous DGE-MRI study of a glioma patient based on T1ρ mapping11 and the in vivo study of Xu et al.9 employing CEST-based DGE-MRI in brain tumor patients. A quantitative evaluation of DGEρ in three regions of interest (ROIs) revealed a substantially increased contrast in the tumor ROIs selected on the DGEρ and GdCE-T1w images compared to normal appearing white matter. Interestingly, the hyperintense tumor areas in the DGEρ images (cf. Fig. 1f–i) partially overlap but still differ from those on the GdCE-T1w image (Fig. 1b). The observed difference in both contrasts is in agreement with the findings of Walker-Samuel et al.3, who did not observe a significant correlation between glucoCEST and GdCE-T1w contrast in an animal study. This allows for the conclusion that DGE-MRI can provide complementary information about pathologies compared to contrast enhanced T1-w MRI, which is the current gold standard method for detecting and characterizing high-grade glioma23 by visualizing blood brain barrier (BBB) disruption. We could not validate whether the enhancing region outside the tumor area (black arrow; Fig. 1g), which was not visible in the native T2-w and GdCE-T1w images (Fig. 1a and b) was an active tumor region or not. Hence, it remains to be shown if DGE-MRI might highlight hidden active regions of the tumor and thus forms a tool for the early detection of cancer. Whereas CEST-based DGE-MRI showed an uptake in blood vessels9, this was not observed by T1p-w DGE-MRI. It remains to be investigated in detail if this is due to the short saturation period of spin-lock compared to CEST or if it has a meaning on the contrast origin level.
Origin of DGEρ contrast
We showed that the DGEp signal increases in tumors. However, the actual origin of the signal changes in DGE-MRI is still under discussion1,2,3,4,5,6,7,8,9,10,11. Chan et al.1 stated that the signal in glucoCEST originates mostly from the extracellular compartment, and, due to lower pH, predominantly from the extracellular-extravascular glucose. Further, Chan et al.1 as well as Walker-Samuel et al.3 showed that FDG-PET and glucoCEST MRI are enhancing similarly. In contrast to Chan et al., Walker-Samuel et al. concluded from the similarity with FDG-PET that also intracellular compounds contribute to the glucoCEST signal. This conclusion was also based on their results of 13C spectroscopy after injection of 13C labeled glucose that showed appearance of glucose, glucose-6-phsophate, fructose phosphates, as well as amino acids such as glutamate, glutamine, taurine and alanine. From phantom experiments they further conclude that glucose and its metabolic products as well as glutamate and glutamine might contribute to the glucoCEST signal, but lactate protons are exchanging too fast to be detectable with CEST3. For the case of glucoCESL, Jin et al.6 also mention the contribution of glucose metabolism products. Thus, it is still under discussion to what extent DGE-MRI is extracellular and consequently only with indirect access to metabolism, or intracellular, which would give more insight to metabolism. From our data, we can only conclude that changes in T1ρ-based DGE-MRI originate from a different compartment than in gadolinium enhanced MRI, which is coherent with both origins, the extracellular extravascular and the intracellular space or a mixture of both. This conclusion is also in coherence with previous publications1,3,5,11. However, in accordance with Jin et al.6 we want to point out that with on-resonant T1ρ-based DGE-MRI all exchanging sites contribute to the signal and, compared to CEST, also the close to water resonating and faster exchanging pools such as lactate have a stronger contribution, as sensitivity of spin-lock to high exchange rates is improved6,17. As the presented technique can track the signal changes fast and accurate, it might become an important tool for further investigations of the origin of the DGE contrast.
Having shown that our contrast is quantitative, we can employ the in vitro calibration to try calculating the corresponding glucose concentration in vivo similar to Jin et al.6. Assuming the relaxivity measured in phantoms (Fig. 3) to be valid also in vivo, the obtained DGEρ or rather change of R1ρ in the tumor would correspond to a glucose concentration increase of up to 40 mM (721 mg/dL), using the relaxivity reported by Jin et al.6 the concentration change would be approximately 25 mM (450 mg/dL). Although Xu et al. measured a venous glucose level of up to 23.7 mM (427 mg/dL) in volunteers about 2–4 min after the injection9, a value between 25 mM and 40 mM still seems to be improbably high. This hints that the observed signal change might not solely originate from the hydroxyl exchange of glucose, but as discussed above, also from glucose metabolic products3,6, and glutamate and glutamine3. Moreover, the relaxivity of the DGE effect potentially differs between the in vivo and in vitro situation as it depends on temperature, pH, and the concentration of exchange catalysts and has not yet been determined directly in vivo or even in tumors.
Unexpected signals and motion correction
After injection, glucose is also expected to rapidly enter the cerebrospinal fluid (CSF) leading to an increase of R1ρ and consequently to a positive DGE contrast in the ventricles as observed in our measurements. However, it has also been reported that a glucose injection results it volumetric changes of the CSF compartments24. This is an explanation for the negative contrast observed in the ventricles by Xu et al. employing CEST-based DGE-MRI9. Such volumetric changes can also lead to a reduction of R1ρ and consequently to a negative DGE contrast employing T1ρ-weighted DGE-MRI explaining the observed signals in the outer CSF compartments, where pixels are expected to be affected by partial volume effects, which most likely result from the limited special resolution in z-direction. Volumetric changes of the CSF lead to an increase of the CSF fraction in the partial volume affected voxels and consequently to negative DGE contrasts, as R1ρ of CSF is about one order of magnitude smaller than for brain tissue8,11. This insight must be included when interpreting DGE uptake of tumors close to CSF regions.
Generally, patient motion is a problem of every contrast based on signal differences between different time points, including all CEST- and CESL-based DGE-MRI approaches, but also dynamic contrast enhanced (DCE) MRI or functional MRI (fMRI). For correction of motion after data acquisition, we employed a rigid registration algorithm. However, for a robust post-process correction of extensive out-of-plane motion the acquisition of an expanded volume is mandatory; for example by applying single-shot 3D MRI sequences such as 3D gradient echo-based MRI25 or echo-planar imaging (EPI) speed-up26, which can easily be combined with the T1ρ-weighted preparation11. An alternative method to reduce patient motion is the application of immobilization devices known from radiation therapy as done by Wang et al.10. In principle, also a combination of post-process motion correction and immobilization of the patient is possible.
In conclusion, dynamic glucose enhanced MRI (DGE-MRI) might open the window to non-invasive observation of glucose uptake and potentially metabolism. Due to its high temporal resolution in combination with a high robustness against field inhomogeneities and a high sensitivity to glucose, T1ρ-weighted DGE-MRI has a high potential to facilitate the translation of glucose enhanced MRI into the clinics. The simple quantitative evaluation can be performed online directly at the scanner to fulfill the clinical demand for practicability. Quantitative DGE further allows a deeper insight into the underlying correlations and in principle enables combined measurements with relaxation affecting contrast agents such as Gd. Further longitudinal studies with larger numbers of patients with different tumor grades are planned to investigate the full potential for detection and staging of cancer or also neurodegenerative diseases by means of the proposed fast and quantitative T1ρ-w DGE-MRI technique.
Methods
R1ρ theory and glucose contrast
For a two-pool system (one water, one solute proton pool) the on-resonant longitudinal relaxation rate in the rotating frame R1ρ is given by27 R1ρ = R2 + Rex where R2 is the transverse relaxation rate of water protons without contributions from chemical exchange and Rex the exchange dependent relaxation rate. Rex can be approximated as6
where Δp is the ratio of concentrations of solute and water protons, kB is the exchange rate (units of s−1) and δ the resonance shift (units of rad/s) between the solute and water proton pools, and ω1 = γB1 is the amplitude of the spin-lock pulse (units of rad/s). For T1ρ-weighted MRI we could show that the difference in signal intensities (ΔS) between a voxel and a reference voxel with different exchange-dependent relaxation, e.g. gray and white brain matter, can be approximated as11
assuming that ΔRex · TSL ≪ 1, where ΔRex is the difference of the exchange dependent relaxation rates between the two voxels and TSL is the spin-lock time. This formula also holds for the same voxel but different time points, e.g. in T1ρ-w DGE-MRI before and after administration of glucose. A similar metric was used by Xu et al. for the evaluation of CEST-based DGE-MRI data5,9.
The dependence of ΔS on R1p and hence on R2 (Eq. 3) indicates that ΔS might be a non-optimal measure for glucose concentration changes in vivo since R2 varies between different tissue types. Dividing equation (3) by the reference signal yields the potentially more robust relative signal change:
Simulations and in vitro measurements
For simulations the Bloch-McConnell equations28 for two pools, one bulk water and one solute pool, were solved numerically as described in Zaiss and Bachert19 using MATLAB (MATLAB R2015b, 2015; The MathWorks Inc., Natick, Massachusetts, USA). The simulation parameters were: R1 = 0.66 s−1, R2 = 20 s−1, δ = 1.5 ppm, kB = 3 kHz, Δp = 9.0·10−4 (≙20 mM), B1 = 5 μT and TSL = 50 ms. For the in vitro measurements two sets of phosphate buffered aqueous solutions (pH ≈ 7.2) with glucose concentrations of 5 mM, 10 mM, 20 mM and 40 mM were used. The relaxation times of the solutions were adjusted by means of 0.095 mM gadoteric acid (Dotarem®; Guerbet, France) and 1.6% agar for the first set of solutions (“agar phantoms”) and 0.45 mM MnCl2 for the second set (“MnCl2 phantoms”).
Patient examination
As part of a clinical study T1ρ-w DGE-MRI was applied in the examination of a 66-year-old male patient with newly diagnosed and histopathologically confirmed glioblastoma (WHO grade IV). The examination was approved by the local ethics committee of the Medical Faculty of the University of Heidelberg and is in accordance with the relevant guidelines and regulations. Written informed consent was received from the patient prior to the examination. The patient was examined after a 6-hour fasting period ensuring a normal blood glucose level before injection. Using an intravenous line 100 ml of 20% D-glucose (SERAG-WIESSNER GmbH & Co. KG, Naila, Germany) were injected manually over 2 min into an arm vein. Two blood samples were taken, one before and the other approximately 25 min after the glucose injection. The blood sugar values, determined by means of a conventional blood sugar meter (Accu-Chek Aviva; Roche Diagnostics, Rotkreuz, Switzerland), were 106 mg/dL (5.9 mM) and 146 mg/dL (8.1 mM) pre- and post-injection, respectively. The complete protocol of the patient examination is sketched in Fig. 5a.
The total examination time, including patient preparation and positioning, morphological and T1ρ-w DGE MRI, as well as the blood glucose measurements, was approximately 60 min. The T1ρ-w DGE-MRI part shown in Fig. 5b consisted of n = 178 individual measurements leading to an acquisition time of about 20 minutes. The first 18 measurements were performed before the start of the glucose injection and yielded the reference for the calculation of the dynamic glucose enhancement (Eq. 1).
Data acquisition and analysis
All MR measurements were performed on a 7T whole-body MR scanner (MAGNETOM 7T, Siemens Healthcare, Erlangen, Germany) using a 24-channel head coil (Nova Medical, Wilmington, MA, USA). The MR sequence used for T1ρ-based MRI consists of an adiabatically prepared spin-lock pulse cluster as described in Schuenke et al.11 and shown in Fig. 5c followed by a conventional MRI readout. The parameters of the adiabatic hypsec-pulses were: RF amplitude B1, max ≥ 20 μT, adiabatic sweep time tadia = 8 ms, bandwidth Δ = 1200 Hz, and μ = 6, where μ is a dimensionless parameter that controls the pulse shape29. The spin-lock frequency was adjusted manually to obtain the desired value of B1 ≈ 5 μT in the region of interest, e.g. the tumor area. For MRI readout we used a centric-reordered single-shot gradient echo (GRE) sequence. In vivo we acquired three axial slices in an interleaved way (matrix = 128 × 104, FoV = 220 × 178 mm2, TE = 3.61 ms, TR = 23 ms, flip angle = 10°, slice thickness = 5 mm, distance factor = 20%). The same MR sequence with an adapted preparation block was used to obtain B1 maps by means of the WASABI30 approach.
In the patient examination we further acquired a stack of 32 high-resolution (0.4 × 0.4 × 2 mm3) T2-weighted images using a Turbo-Spin-Echo (TSE) sequence (TE = 52 ms, TR = 12340 ms). The Gadolinium contrast-enhanced T1-weighted (GdCE-T1w) images were acquired 10 days prior to the 7T examination in the course of a clinical MR protocol at 3T. The GdCE-T1w, T2-w and T1ρ-w images were co-registered and the slice thickness of the GdCE-T1w and T2-w images was interpolated to the slice thickness (5 mm) of the T1ρ-w images using a multi modal rigid registration algorithm in the DKFZ Image Processing Platform - an in-house version of the Medical Imaging Interaction Toolkit (MITK)31. All further post-processing and data analysis, including a rigid in-plane motion correction of the T1ρ-w images was performed using self-written software in MATLAB. All errors were calculated taking into account the law of error propagation.
Additional Information
How to cite this article: Schuenke, P. et al. Fast and Quantitative T1ρ-weighted Dynamic Glucose Enhanced MRI. Sci. Rep. 7, 42093; doi: 10.1038/srep42093 (2017).
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
Chan, K. W. Y. et al. Natural D-glucose as a biodegradable MRI contrast agent for detecting cancer. Magn. Reson. Med. 68, 1764–73 (2012).
Rivlin, M., Horev, J., Tsarfaty, I. & Navon, G. Molecular imaging of tumors and metastases using chemical exchange saturation transfer (CEST) MRI. Sci. Rep. 3, 3045 (2013).
Walker-Samuel, S. et al. In vivo imaging of glucose uptake and metabolism in tumors. Nat. Med. 19, 1067–72 (2013).
Nasrallah, F. A., Pagès, G., Kuchel, P. W., Golay, X. & Chuang, K.-H. Imaging brain deoxyglucose uptake and metabolism by glucoCEST MRI. J. Cereb. Blood Flow Metab. 33, 1270–8 (2013).
Xu, X. et al. Dynamic glucose enhanced (DGE) MRI for combined imaging of blood-brain barrier break down and increased blood volume in brain cancer. Magn. Reson. Med. 74, 1556–63 (2015).
Jin, T., Mehrens, H., Hendrich, K. S. & Kim, S.-G. Mapping brain glucose uptake with chemical exchange-sensitive spin-lock magnetic resonance imaging. J. Cereb. Blood Flow Metab. 34, 1402–10 (2014).
Zu, Z., Spear, J., Li, H., Xu, J. & Gore, J. C. Measurement of regional cerebral glucose uptake by magnetic resonance spin-lock imaging. Magn. Reson. Imaging 32, 1078–84 (2014).
Jin, T., Mehrens, H., Wang, P. & Kim, S.-G. Glucose metabolism-weighted imaging with chemical exchange-sensitive MRI of 2-deoxyglucose (2DG) in brain: Sensitivity and biological sources. Neuroimage 143, 82–90 (2016).
Xu, X. et al. Dynamic Glucose-Enhanced (DGE) MRI: Translation to Human Scanning and First Results in Glioma Patients. Tomogr. a J. imaging Res. 1, 105–114 (2015).
Wang, J. et al. Magnetic Resonance Imaging of Glucose Uptake and Metabolism in Patients with Head and Neck Cancer. Sci. Rep. 6, 30618 (2016).
Schuenke, P. et al. Adiabatically prepared spin-lock approach for T1ρ-based dynamic glucose enhanced MRI at ultrahigh fields. Magn. Reson. Med., doi: 10.1002/mrm.26370 (2016).
Tofts, P. S. Modeling tracer kinetics in dynamic Gd-DTPA MR imaging. J. Magn. Reson. Imaging 7, 91–101 (1997).
Duvvuri, U. et al. Human knee: in vivo T1(rho)-weighted MR imaging at 1.5T–preliminary experience. Radiology 220, 822–6 (2001).
Akella, S. V. S. et al. Proteoglycan-induced changes in T1rho-relaxation of articular cartilage at 4T. Magn. Reson. Med. 46, 419–23 (2001).
Regatte, R. R., Akella, S. V. S., Borthakur, A., Kneeland, J. B. & Reddy, R. Proteoglycan depletion-induced changes in transverse relaxation maps of cartilage: comparison of T2 and T1rho. Acad. Radiol. 9, 1388–94 (2002).
Singh, A. et al. High resolution T1ρ mapping of in vivo human knee cartilage at 7T. PLoS One 9, e97486 (2014).
Jin, T. & Kim, S.-G. Advantages of chemical exchange-sensitive spin-lock (CESL) over chemical exchange saturation transfer (CEST) for hydroxyl- and amine-water proton exchange studies. NMR Biomed. 27, 1313–24 (2014).
Jin, T., Autio, J., Obata, T. & Kim, S.-G. Spin-locking versus chemical exchange saturation transfer MRI for investigating chemical exchange process between water and labile metabolite protons. Magn. Reson. Med. 65, 1448–60 (2011).
Zaiss, M. & Bachert, P. Chemical exchange saturation transfer (CEST) and MR Z-spectroscopy in vivo: a review of theoretical approaches and methods. Phys. Med. Biol. 58, R221–69 (2013).
Zaiss, M. et al. Relaxation-compensated CEST-MRI of the human brain at 7T: Unbiased insight into NOE and amide signal changes in human glioblastoma. Neuroimage 112, 180–8 (2015).
Yadav, N. N. et al. Natural D-glucose as a biodegradable MRI relaxation agent. Magn. Reson. Med. 72, 823–8 (2014).
Zaiss, M. et al. Inverse Z-spectrum analysis for spillover-, MT-, and T1 -corrected steady-state pulsed CEST-MRI–application to pH-weighted MRI of acute stroke. NMR Biomed. 27, 240–52 (2014).
Dhermain, F. G., Hau, P., Lanfermann, H., Jacobs, A. H. & van den Bent, M. J. Advanced MRI and PET imaging for assessment of treatment response in patients with gliomas. Lancet. Neurol. 9, 906–20 (2010).
Puri, B. K., Lewis, H. J., Saeed, N. & Davey, N. J. Volumetric change of the lateral ventricles in the human brain following glucose loading. Exp. Physiol. 84, 223–6 (1999).
Krishnamoorthy, G., Nanga, R. P. R., Bagga, P., Hariharan, H. & Reddy, R. High quality three-dimensional gagCEST imaging of in vivo human knee cartilage at 7 tesla. Magn. Reson. Med., doi: 10.1002/mrm.26265 (2016).
Jones, C. K. et al. In vivo three-dimensional whole-brain pulsed steady-state chemical exchange saturation transfer at 7T. Magn. Reson. Med. 67, 1579–89 (2012).
Trott, O. & Palmer, A. G. R1rho relaxation outside of the fast-exchange limit. J. Magn. Reson. 154, 157–60 (2002).
McConnell, H. M. Reaction Rates by Nuclear Magnetic Resonance. J. Chem. Phys. 28, 430 (1958).
Silver, Joseph & Hoult Selective spin inversion in nuclear magnetic resonance and coherent optics through an exact solution of the Bloch-Riccati equation. Phys. Rev. A, Gen. Phys. 31, 2753–2755 (1985).
Schuenke, P. et al. Simultaneous mapping of water shift and B1 (WASABI)-Application to field-Inhomogeneity correction of CEST MRI data. Magn. Reson. Med. 77 (2), 571–580 doi: 10.1002/mrm.26133 (2017).
Stein, D., Fritzsche, K. H., Nolden, M., Meinzer, H. P. & Wolf, I. The extensible open-source rigid and affine image registration module of the Medical Imaging Interaction Toolkit (MITK). Comput. Methods Programs Biomed. 100, 79–86 (2010).
Acknowledgements
We thank all collaborators from the German Cancer Research Center and the University Hospital Heidelberg involved in this interdisciplinary project for their support and valuable contributions. We have no funding to report.
Author information
Authors and Affiliations
Contributions
P.S. Study design, sequence and software development, data acquisition, analysis and interpretation, graph and manuscript preparation. D.P. study design, patient enrollment, data acquisition and interpretation, manuscript editing. C.K. patient enrollment, data acquisition and interpretation, manuscript editing. J.W. data acquisition and interpretation, manuscript editing. P.B. study design and supervision, data interpretation, manuscript editing. M.E.L., H.P.S. and A.R. study design and supervision. M.Z. Study design, sequence development, data analysis and interpretation, manuscript editing. All authors read and approved the manuscript.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing financial interests.
Rights and permissions
This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/
About this article
Cite this article
Schuenke, P., Paech, D., Koehler, C. et al. Fast and Quantitative T1ρ-weighted Dynamic Glucose Enhanced MRI. Sci Rep 7, 42093 (2017). https://doi.org/10.1038/srep42093
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/srep42093
- Springer Nature Limited
This article is cited by
-
GlucoCEST MRI for the Evaluation Response to Chemotherapeutic and Metabolic Treatments in a Murine Triple-Negative Breast Cancer: A Comparison with[18F]F-FDG-PET
Molecular Imaging and Biology (2022)
-
What do we know about dynamic glucose-enhanced (DGE) MRI and how close is it to the clinics? Horizon 2020 GLINT consortium report
Magnetic Resonance Materials in Physics, Biology and Medicine (2022)
-
GLINT: GlucoCEST in neoplastic tumors at 3 T—clinical results of GlucoCEST in gliomas
Magnetic Resonance Materials in Physics, Biology and Medicine (2022)
-
„Chemical exchange saturation transfer“ (CEST)
Der Radiologe (2021)
-
The beginning of the end for conventional RECIST — novel therapies require novel imaging approaches
Nature Reviews Clinical Oncology (2019)