What do we know about dynamic glucose-enhanced (DGE) MRI and how close is it to the clinics? Horizon 2020 GLINT consortium report

Abstract

Cancer is one of the most devastating diseases that the world is currently facing, accounting for 10 million deaths in 2020 (WHO). In the last two decades, advanced medical imaging has played an ever more important role in the early detection of the disease, as it increases the chances of survival and the potential for full recovery. To date, dynamic glucose-enhanced (DGE) MRI using glucose-based chemical exchange saturation transfer (glucoCEST) has demonstrated the sensitivity to detect both D-glucose and glucose analogs, such as 3-oxy-methyl-D-glucose (3OMG) uptake in tumors. As one of the recent international efforts aiming at pushing the boundaries of translation of the DGE MRI technique into clinical practice, a multidisciplinary team of eight partners came together to form the "glucoCEST Imaging of Neoplastic Tumors (GLINT)" consortium, funded by the Horizon 2020 European Commission. This paper summarizes the progress made to date both by these groups and others in increasing our knowledge of the underlying mechanisms related to this technique as well as translating it into clinical practice.

Keywords: 3-Oxy-methyl-D-glucose; CEST; Cancer; DGE MRI; Glucose; MRI; glucoCEST.

Fig. 1

A 2PLSM image of neuronal expression of FLIIP, B 2PLSM image of astrocytic expression of FLIIP, C Experimental setup of simultaneous DGE MRI (7 T Bruker BioSpec 70/30) and fiber photometry measurements: (a) optic fiber connector, (b) dichroic mirror 455 nm, (c) dichroic mirror 515 nm, (d) bandpass filter 530/43 nm, (e) bandpass filter 475/42 nm

Fig. 2

Glucose response curves upon intravenous injection of 120 μl 50% w/v glucose solution at normoglycemia after 10 min baseline. A DGE signal, whole brain, 1 mm slice, B 2PLSM (λ_exc = 870 nm) in astrocytes and neurons, respectively

Fig. 3

Representative T₂w images (A), DGE map after d-glucose i.v. injection (B), CEST contrast (C) and tumor pH maps (D) after iopamidol I.v. injection, fused ¹⁸F-FDG-PET/CT images (E) for 4T1 and PC3 tumor-bearing mice. Average values calculated for each tumor model of Glucose ΔST% (F), ¹⁸F-FDG PET uptake as %ID/cc (G) and tumor pH (H)

Fig. 4

Bar graph showing % of MTR_asym of 20 mM glucose solution (10% D₂O) at several PBS concentrations at the typical frequency offsets of the hydroxyl peaks (A) 1.3 ppm, (B) 2.1 ppm and (C) 2.88 ppm from the water peak (T = 37 °C, 11.7 T). ¹H NMR spectra of the hydroxyl protons of 0.1 M D-Glc solution (T = 4 °C, pH = 5.4, 11.7 T). Spectra were recorded on a fresh sample (D) and several hours after the sample preparation (E). From [44], with permission

Fig. 5

3OMG-based DGE MRI and ¹⁸F-FDG-PET/CT images from five tumors of a murine model (4T1 cells). A A coronal view of an anatomical T₂-weighted MR images (7 T field) before 3OMG administration showing the tumor (green arrow) and the urinary bladder (red arrow). B % CEST images 60 min after per os administration with 3OMG, 1.0 g/kg (at a frequency offset of 1.2 ppm, B₁ = 2.4 µT). A significant CEST contrast was obtained in the tumor and the urinary bladder as well as areas suspected to be metastases. C ¹⁸F-FDG-PET/CT coronal view obtained 60 min after IV injection of ¹⁸FDG showing accumulation mainly in the tumor (green arrow) and urinary bladder (red arrow). D Correlation between 3OMG % CEST contrast and % ID/mL value in the five tumors from a murine model ± S.D [44]. From [44], with permission

Fig. 6

A ¹H‐decoupled ¹³C NMR spectra of 100 mM 3OMG solution (a), 100 mM [6‐¹³C]‐3OMG solution (b), combined extracts from 4T1 tumors treated with [6‐¹³C]‐3OMG (1 g/kg) (c), combined untreated extracts from 4T1 tumors (d), enlargement of (c and e). The arrow represents the expected position of the phosphorylated product of 3OMG (3OMG‐6‐phosphate) on the basis of the observed glucose‐6‐phosphate at 65.6 ppm and the 2‐DG‐6‐phosphate peak at 66.0 ppm. B ¹H‐decoupled ³¹P NMR spectra of combined extracts from 4T1 tumors treated with [6‐¹³C]‐3OMG (1 g/kg) (a), combined untreated extracts from 4T1 tumors (b). GPC‐glycerophosphocholine, GPE‐glycerophosphoethanolamine, Pi‐inorganic phosphate; spectra were calibrated according to GPC (0.49 ppm). From [46], with permission

Fig. 7

In vivo 2OMG and 6DG CEST MRI measurements in 4T1 tumors (7 T field). A, C T₂-RARE anatomical images before administration of the agents. B, D MTR_asym images at 1.0 and 1.2 ppm following treatment with 2OMG (3 g/kg, IP) and 6DG (2.0 g/kg, IP), respectively, overlaid onto the T₂ anatomical image. From [44], with permission

Fig. 8

A Simultaneous multi‐B₁‐pH‐fit of 25 Z‐spectra of 20 mM glucose model solutions acquired at 14.1 T yields glucose hydroxyl exchange rates as a function of pH at T = 37 °C B and R_2A and anomeric ratio C. From [48], with permission

Fig. 9

Optimization of the HSExp pulse for 3 T. A–C SSD to an analytical T_1ρ spectrum, which were used for pulse optimization [22]. Parameters resulting in a minimal SSD were ∆f = 2.5 kHz, µ = 65, and t_window = 3.5 ms, marked with a red square. D The SL cluster at 0 ppm for a TSL = 120 ms. E Measured Z‐spectra and their standard error for this pulse cluster in a WM and GM ROI. DGE_ρ offsets acquired in the dynamic experiment are marked with a dashed line (0.6, 0.9, 1.2, and 1.5 ppm). Simulated DGE_ρ effect after d-glucose injection (F–H) in the steady‐state CEST regime (F), the intermediate regime with only one second of saturation (G), and the SL regime with 100 ms of saturation (H). From [22] and [48], with permission

Fig. 10

Effect of motion correction, interleaved M₀, and dynamic B₀ correction. A ∆DGE_ρ [%] map of a volunteer measurement (image number 16 at ∆ω = 0.9 ppm) after motion correction. Here, the entire left hemisphere seems to be affected by a strong hypointensity. B ∆DGE_ρ map with a normalization to the corresponding image at − 300 ppm. Here, the image is more flat, but still shows slight correlation with B₀ (D), especially in the anterior part. C ∆DGE_ρ map with the same normalization as (B) and an additional B₀ correction as proposed in Windschuh et al. [68]. The dynamic B₀ correction normalizes the anterior part that is hypointense in (B). D ∆B₀ map [ppm] relative to the beginning of the measurement. E Histogram for the images in (A–C). From [22], with permission

Fig. 11

A MTR_asym signal integrated in the range of 2—3 ppm before B₀ and B₁ correction shows field drifts both in tumor and contralateral regions of a patient with prostate cancer. It is worthwhile to note that the changes due to B₀ drift are much larger in the body, due to the increased drift observed. In this case the B₀ drifts across slice and entire scan duration were found to be 25 Hz (0.2 ppm) and 200 Hz (1.56 ppm), respectively. B After B₀ correction, no significant enhancement in MTR_asym signal is observed and the signal intensity is significantly reduced. From [25], with permission

Fig. 12

T₂ FLAIR (A, E, I), T₁‐ce (B,F,J), T₁‐ce with overlaid ∆DGE_ρ map (C, G, K), and the ∆DGE_ρ‐maps (D,H,L) of patient 1 (A–D, ~ 7 min post-injection), 2 (E–H, ~ 7 min post-injection), and 3 (I–L, ~ 9 min post-injection). From [22], with permission