极T代谢磁共振全球科研集锦

45

Wang et al

Radiology: Volume 291: Number 2—May 2019 n radiology.rsna.org 277

selectively suppress the 13C signal from normal hepatocytes (70),

as compared with other cell types such as inflammatory cells.

Kidney Disease

In the kidneys, HP 13C MRI has been used to investigate metabolic

changes related to hypoxia and oxidative stress (17,71–74),

two key factors implicated in progressive kidney injury.

Increased renal [1-13C]pyruvate–to-lactate conversion was observed in models of diabetes and has been hypothesized to be related to intrarenal pseudohypoxia from hyperglycemia (71,72).

In models of acute kidney injury (ischemia-reperfusion injury),

decreased [1-13C]pyruvate–to-bicarbonate conversion has been

noted, corresponding to decreased PDH activity and mitochondrial energy production (73,74). !e [1-13C]pyruvate–to-

(63–69). For example, increased [1-13C]pyruvate–to-alanineand-lactate conversion was observed in a model of chemically

induced inflammatory liver injury (65), and decreased [1-13C]

DHA–to–vitamin C conversion was noted in a model of dietinduced steatohepatitis (18). !ese results suggest that such an

imaging approach may address a currently unmet clinical need

for noninvasive diagnosis and treatment monitoring of nonalcoholic steatohepatitis (NASH), a disease of growing concern

in the United States and worldwide. HP 13C dihydroxyacetone,

a gluconeogenesis precursor, has been used to probe hepatic energy metabolism (68,69), the alteration of which is implicated in

both NASH and cirrhosis. To increase the specificity of detected

13C metabolism, a clinically used liver-targeted contrast agent (ie,

gadoxetate) can be administered prior to 13C probe injection to

Figure 3: Hyperpolarized (HP) carbon 13 (13C) MRI in a transgenic adenocarcinoma mouse model of prostate cancer. A,

Representative hematoxylin-eosin–stained pathology sections and HP 13C spectra after injection of HP [1-13C]pyruvate in a

normal mouse prostate, an early-stage prostate tumor, a late-stage prostate tumor, and a lymph node metastasis. At histologic

examination, normal murine prostate was glandular, with secretory epithelial cells lining the glands (arrowhead). In prostate tumors, there was gradual replacement of the secretory epithelial cells by less differentiated epithelial cells, until the glands were

completely eliminated and only anaplastic sheets of pleomorphic cells with irregular nuclei remained in the late stage tumors

(arrow). The 13C spectra show an increase in 13C lactate and 13C lactate/pyruvate ratio in late-stage tumor and nodal metastasis. B, Axial T2-weighted anatomic MR image and overlay of HP 13C lactate image on T2-weighted image show a qualitatively

high level of lactate in a late-stage tumor. Units for color bar = arbitrary units of signal intensity. C, Boxplots show quantitative 13C lactate signal in the four histologically defined groups, with late-stage tumors having significantly higher lactate than earlystage tumors. SNR = signal-to-noise ratio. (Adapted from reference 23.)

极T代谢磁共振全球科研集锦

46

State of the Art and Future Directions of Hyperpolarized Carbon 13 MRI

278 radiology.rsna.org n Radiology: Volume 291: Number 2—May 2019

clinical polarizers can achieve up to 50% polarization of [1-13C]

pyruvate, more than a twofold increase compared with the initial

prototype polarizer used for the phase I human study.

Fast Imaging Techniques

!e rapid metabolism and relatively short T1 relaxation times

of most biologically relevant HP 13C probes have required the

development of novel multichannel coil array systems, as well

as specialized rapid pulse sequences, and many of these are now

commercially available. Multichannel 13C radiofrequency coil

arrays allow increased sensitivity and the use of rapid imaging

techniques including parallel imaging. For abdominal studies, a

commercial 16-channel torso array is available. For other applications, a variety of both custom and commercial 13C MRI coils

have been developed with both single-channel receive and multichannel arrays of up to 32 channels for brain tumor studies.

In terms of pulse sequences, two general approaches can be

used: a spectroscopic technique and an imaging-based technique. Rapid spectroscopic techniques, such as echo-planar

spectroscopic imaging (EPSI), generate a 13C spectrum for every

voxel that is imaged (83). !ese sequences are useful when the

exact metabolites are not known, when there are numerous metabolites, or when inhomogeneity in the magnetic field causes

the spectral frequencies to be shifted across the sample. In an

alternative image-based approach, each metabolite is excited in

turn and then imaged by using a conventional imaging sequence

such as echo-planar imaging (84), spiral MRI (85), or balanced

steady-state free precession (86). !e advantage of this approach

is that it is faster than EPSI and directly generates images on

the MRI unit, which facilitates clinical interpretation. !e disadvantages of the imaging-based approaches are that they require

the metabolite frequencies to be known beforehand and they are

sensitive to nonuniform magnetic fields. To improve the efficient

use of HP signal as well as the accuracy and robustness of HP

MRI, bolus tracking and real-time B1

calibration methods have

also been implemented recently (87).

Because many metabolic processes are rapid, fast dynamic

HP 13C imaging of the substrates and their metabolites is advantageous (88). Dynamic 13C acquisition enables obtaining quantitative metabolic parameters such as the apparent rate constant

for pyruvate-to-lactate conversion (kPL), which is less dependent

on the precise acquisition and contrast bolus timing. Because the

HP 13C signal is nonrenewable, specialized fast imaging strategies

are used to conserve magnetization and to allow dynamic data

acquisition at multiple time points. For example, compressed

sensing reconstruction can be used to minimize the amount of

data required. Additionally, special radiofrequency pulses can be

used to apply different flip angles to the HP 13C substrates and

their metabolites to maximize the signal from the metabolites

while preserving the magnetization of the substrates (89).

!e MRI unit field strength affects the T1 relaxation time of

HP probes. For example, HP 13C pyruvate has slightly shorter

T1 at 3.0 T than at 1.5 T. However, the chemical shift separation between pyruvate and its metabolites in hertz at 3.0 T is

twice that at 1.5 T; this enables better metabolic quantification.

!erefore, current clinical studies of HP 13C pyruvate MRI are

commonly performed with 3.0-T MRI units.

lactate conversion in the injured kidney changes dynamically

during the evolution of acute kidney injury (73,74) and may

potentially yield information on renal tubular injury during

the progression from acute kidney injury to chronic kidney

disease. Noninvasive predictors of such progressive kidney

injuries are urgently needed given the increasing incidence of

chronic kidney disease. !e redox probe HP [1-13C]DHA has

been shown to sensitively monitor the level of glutathione, a

major antioxidant, during progression of diabetes and following treatment targeted to oxidative stress (17). Additionally,

the corticomedullary gradient of urea, which is a key osmolyte

in the kidney’s ability to concentrate urine, can be monitored

by using HP 13C urea (75,76), suggesting the potential of this

probe to monitor kidney function. While concerns about toxicity have limited the use of gadolinium-based contrast agents

in patients with renal insufficiency, HP MRI with endogenous

probes such as pyruvate and urea is particularly promising for

the noninvasive diagnosis and monitoring of kidney disease.

Inflammation

Chronic inflammation is increasingly recognized as a common

factor in many diseases and as a target for therapy. Inflammatory tissues have Warburg-like metabolism and hence can be

interrogated by using HP pyruvate MRI. In addition to liver

inflammation (as discussed above), increased HP [1-13C]pyruvate–to-lactate conversion has been shown to reflect inflammation in models of radiation-induced lung injury (77), neuroinflammation (78), and inflammatory arthritis (79). Response to

anti-inflammatory treatment can similarly be monitored with

HP [1-13C]pyruvate MRI (80). Additionally, recent work has

explored new HP probes, such as HP [6-13C]arginine (81), that

specifically target inflammatory cell metabolism. !is is of particular interest in cancer treatment monitoring, as HP [1-13C]

pyruvate cannot help differentiate lactate production from cancer cells from that associated with inflammatory cells.

Technical Advances Enabling Clinical

Translation of HP 13C MRI

!e true potential of HP 13C MRI lies in its rapid progress in

clinical translation for patient care. Several technical advances

have been made in recent years to achieve this goal. Figure 4

shows the requisite components of human HP 13C studies.

!ese components are explained further below.

Clinical Polarizer

Clinical polarizers are now commercially available (SPINlab; GE

Research Circle Technology, Waukesha, Wis) for patient studies

(82), with 23 sited around the world. !ese clinical polarizers allow up to four samples to be polarized at the same time, permitting

the injection of multiple probes during a single examination. !e

preparation of the probes uses a “pharmacy kit,” which is filled either in a licensed pharmacy in a sterile compounding environment

or by using Good Manufacturing Practice with terminal sterilization following the polarization. !e clinical polarizer contains a

quality-control module that measures pH, pyruvate and residual

radical concentration, polarization, and sample temperature prior

to injection to ensure the probe is safe for clinical use. Current

极T代谢磁共振全球科研集锦

47

Wang et al

Radiology: Volume 291: Number 2—May 2019 n radiology.rsna.org 279

kilogram of body weight was administered.

Despite being a supra-physiologic level of

pyruvate, this dose is rapidly metabolized,

and no significant adverse events have been

reported to date in clinical studies that

used the same dose. In the initial human

study, elevated [1-13C]pyruvate–to-lactate

conversion was demonstrated in the tumor

compared with adjacent normal prostate

tissue. In one study subject, elevated lactate

signal was seen in a focus of tumor (later

confirmed at biopsy) that was not visible

on either the T2-weighted or the diffusionweighted images. "ese results highlight

the potential of this technology to improve

tumor detection and characterization. "e

first results demonstrating metabolic response to androgen deprivation therapy by

using HP [1-13C]pyruvate MRI have also

been described (93) (Fig 5). After 6 weeks

of treatment, there was complete abrogation of the HP 13C lactate signal while there

was no change on T2-weighted images and

only a modest change on apparent diffusion

coefficient maps, illustrating the potential of HP lactate as a

biomarker of treatment response. Preliminary data have also

demonstrated that HP [1-13C]pyruvate MRI provides reproducible measurements of metabolic changes in the prostates of

patients who underwent repeated injections of [1-13C]pyruvate

(94). Several other clinical investigations in prostate cancer are

currently ongoing, including studies to predict tumor grade in

patients before prostatectomy and to assess early treatment

response in men with metastatic prostate cancer.

Brain Tumors

In an initial study of eight subjects with a diagnosis of primary

brain tumors, HP [1-13C]pyruvate was rapidly transported across

the blood-brain barrier and converted to the metabolites lactate

and bicarbonate (95) (Fig 6). "e non-tumor brain showed high

bicarbonate signal that was not detected in the tumors, suggesting that this technique may be valuable for probing mitochondrial oxidative metabolism and its alteration in brain tumors.

Another recent study (96) similarly confirmed the feasibility of

acquiring high-quality metabolite maps in patients with primary

or metastatic brain tumors. "ese initial clinical studies demonstrated successful technical implementation of HP [1-13C]pyruvate MRI for brain metabolism evaluation. While additional

technical optimization is required, the data support further investigation of the clinical utility of this technology in patients

with brain tumors, as well as in patients with other neurologic

diseases.

Cardiac Metabolism

"e first images of HP 13C MRI in the human heart were reported in a pilot study of healthy volunteers (97). "e metabolism of [1-13C]pyruvate and the generation of bicarbonate

signal in the left ventricular myocardium were visualized with

Data Analysis

Because HP 13C MRI signal changes rapidly over the course of

a study, new approaches for data analysis and quantification are

required. One common approach is to express the total signal

of downstream metabolites (ie, lactate, alanine) as a fraction of

the signal from the injected probe (ie, pyruvate). Alternatively,

pharmacokinetic models have been proposed to quantify the

apparent rate constants for substrate conversions (2,90–92).

For example, kPL can be calculated in patients (92). "ese approaches need to be further tested and refined in clinical studies to establish the most accurate and reproducible quantitative

analysis methods for HP MRI.

Clinical HP Investigations

"e two probes [1-13C]pyruvate and [2-13C]pyruvate are currently the probes that have received regulatory approval in the

United States (U.S. Food and Drug Administration Investigational New Drug) for clinical investigation. Around the world,

10 sites have performed HP pyruvate MRI in more than 200

human subjects, with several other sites planning clinical studies in the near future. Table E1 (online) lists the up-to-date

clinical trials that have registered with ClinicalTrials.Gov. "ese

trials are predominantly oncology focused but also include investigations in heart and liver disease and traumatic brain injury, suggesting the potential of this technology for metabolic

imaging in a range of human diseases.

Prostate Cancer

"e first human study of HP [1-13C]pyruvate MRI was performed in study subjects with localized prostate cancer at the

University of California San Francisco and established the

safety and feasibility of this new metabolic imaging technique

(2). In that study, a dose of up to 0.43 mL [1-13C]pyruvate per

Figure 4: Schematic shows the required components for clinical hyperpolarized carbon

13 (13C) MRI studies. The pharmacy kits used for preparing sterile 13C probes, the clinical

polarizers with built-in quality-control units, and specialized MRI detector hardware are now

commercially available. Many of the fast 13C pulse sequences and postprocessing tools

are available as open source. In the case of [1-13C]pyruvate, the Investigational New Drug

resource for its use is available from the National Cancer Institute to assist sites in obtaining

U.S. Food and Drug Administration (FDA) regulatory approval for clinical HP MRI studies.

kPL = apparent rate constant for pyruvate-to-lactate conversion.

极T代谢磁共振全球科研集锦

48

State of the Art and Future Directions of Hyperpolarized Carbon 13 MRI

280 radiology.rsna.org n Radiology: Volume 291: Number 2—May 2019

ney, and liver disease are also promising areas for exploration in

humans.

Comparison with PET

Although FDG PET and HP pyruvate MRI are both sensitive

to the metabolic changes that occur in tumors, there are

important distinctions that make the two methods complementary. As already discussed, HP MRI with [1-13C]pyruvate provides information at a key branch point of metabolic pathways

involved in downstream glucose metabolism—information that

is not accessible to FDG PET. For example, prostate cancer often has similar FDG avidity to the background prostate (99). In

contrast, a grade-dependent increase in HP pyruvate-to-lactate

conversion has been observed in a TRAMP model, as well in

patient-derived prostate tissues (23,100), making HP pyruvate

MRI a promising tool for assessing tumor aggressiveness. Additionally, FDG PET is sensitive to both glucose uptake and

the ability of the cell to trap FDG within cells. For example,

some hepatocellular carcinomas have low expression of the glucose transporter GLUT1 but high expression of glucose 6 phosphatase, which phosphorylates FDG and traps it intracellularly,

thereby altering the FDG signal (101). Furthermore, because all

PET agents produce photons with the same energy, the administered PET probes cannot be distinguished from their products,

and it is difficult to image several different PET probes simultahigh signal-to-noise ratio. More recently, the same investigators

also demonstrated the feasibility of using HP [1-13C]pyruvate to

detect metabolic alterations in study subjects with hypertrophic

cardiomyopathy (98). Preliminary data showed significantly

elevated bicarbonate signal near the cardiac apex, corresponding to the known location of disease. #ere was also a different

spatial distribution of the bicarbonate signal in cardiomyopathy

compared with the normal heart. #ese initial data suggest the

exciting possibilities of imaging altered cardiac energetics in patients with heart disease and potentially improving diagnosis and

treatment monitoring in these patients.

Other Emerging Applications

In addition to the prostate, brain, and cardiac applications described above, preliminary human results have been obtained

in the liver (Fig 7), kidney, pancreas, breast, and bone. As has

already been suggested in initial patient data (93), the ability to

rapidly monitor response to treatment is likely one of the most

promising applications of HP 13C MRI. Particularly in the context of cancer, early detection of nonresponse is essential for

minimizing toxicities and redirecting to more effective therapies.

Other potential oncologic applications include risk stratification

of tumors after the initial diagnosis and determining the most

metabolically active region of tumors to target tissue sampling.

Furthermore, metabolic changes occurring with diabetes, kidFigure 5: Representative axial T2-weighted MR image, water apparent diffusion coefficient (ADC) image, T2-weighted MR image with overlaid

apparent rate constant for pyruvate (Pyr)-to-lactate (Lac) conversion (kPL), and corresponding hyperpolarized (HP) carbon 13 (

13C) spectral array in

a 52-year-old patient with extensive high-grade (Gleason 4 + 5) prostate cancer, A, before therapy and, B, 6 weeks after initiation of androgen

ablation and chemotherapy. Before treatment, the region of prostate cancer can be seen (arrows) as low signal on the T2-weighted and ADC images, with high HP lactate seen in the spectral array (vertical axis = arbitrary signal intensity units; horizontal axis = frequency in parts per million)

and high kPL. At 6 weeks after initiation of androgen deprivation therapy, there was near complete abrogation of elevated HP lactate peaks on the

spectral array and associated marked reduction in tumor kPL (maximum kPL, 0.025 sec21 at baseline and 0.007 sec21 at follow-up). Notably, there

was negligible change in the size of tumor on T2-weighted MR images and only a modest change on ADC images, supporting the utility of HP

[1-13C]pyruvate MRI in detecting early metabolic responses prior to changes on conventional images. Concordant with these findings, the patient

subsequently achieved a marked clinical response, with an undetectable serum prostate-specific antigen level 6 months after treatment initiation.

(Reprinted, with permission, from reference 91.)

极T代谢磁共振全球科研集锦

49

Wang et al

Radiology: Volume 291: Number 2—May 2019 n radiology.rsna.org 281

context, the advent of the integrated PET/MRI units facilitate

such multiparametric examinations.

Challenges and Future Directions

For HP 13C MRI to be adopted as a clinical molecular imaging

tool to advance patient care, further efforts are needed in sevneously. In contrast, because HP 13C probes and their metabolites have different resonance frequencies, they can be imaged

simultaneously to provide information on multiple metabolic

processes (13). Finally, because PET and HP 13C MRI depict

complementary aspects of tumor metabolism, it is attractive to

explore the combined performance of both modalities. In that

Figure 6: Hyperpolarized (HP) [1-13C]pyruvate MR images in two patients with treated glioblastoma multiforme. Patient A (top row)

had progressing tumor at the time of the HP study, while patient B (bottom row) had stable tumor at the time of the HP study. HP [1- 13C]pyruvate was rapidly transported across the blood-brain barrier and converted to the metabolites lactate and bicarbonate. There

was substantial pyruvate-to-lactate and pyruvate-to-bicarbonate conversion in the normal brain. The tumor (arrows) in patient A, who

had progressive tumor, showed moderate pyruvate-to-lactate conversion when compared with normal brain. The tumor (arrow) in patient B, who had clinically stable tumor, showed no substantial pyruvate-to-lactate conversion. Both tumors showed a lack of pyruvateto-bicarbonate conversion, suggesting reduced mitochondrial oxidative metabolism. Color bars = metabolite ratios. The differences in

the scales of the color bars are due to the single-band constant flip angle excitation scheme used for patient A and the multiband variable flip angle excitation scheme used for patient B. These initial findings support the use of HP [1-13C]pyruvate MRI for investigating

metabolic reprogramming in brain tumors. FLAIR = fluid-attenuated inversion recovery, Gad = gadolinium. (Reprinted from reference

93.)

Figure 7: Hyperpolarized (HP) [1-13C]pyruvate MR images in a patient with breast cancer metastatic to the liver. The study

was performed by using a 16-channel abdomen carbon 13 (13C) coil array permitting coverage of the whole upper abdomen

and a dynamic 13C echo-planar spectroscopic imaging technique at a 2-second time resolution. T2-weighted single-shot fast

spin-echo (SSFSE) image (left) shows multiple liver metastases. The kPL (the apparent rate constant for pyruvate-to-lactate conversion in sec21) map overlaid on the T2-weighted SSFSE image (right) shows liver metastases (arrows) with elevated pyruvate-tolactate conversion, consistent with metabolically active tumors.

极T代谢磁共振全球科研集锦

50

State of the Art and Future Directions of Hyperpolarized Carbon 13 MRI

282 radiology.rsna.org n Radiology: Volume 291: Number 2—May 2019

formulation of new probes, quantification and standardization

of results, and multicenter clinical trials.

Disclosures of Conflicts of Interest: Z.J.W. Activities related to the present article: disclosed no relevant relationships. Activities not related to the present

article: holds stock or stock options in Nextrast. Other relationships: disclosed no

relevant relationships. M.A.O. disclosed no relevant relationships. P.E.Z.L. Activities

related to the present article: institution received a grant from GE Healthcare. Activities not related to the present article: is on the speakers bureau of GE Healthcare;

receives royalties from GE Healthcare, Philips, and Siemens. Other relationships:

receives royalties from patents licensed to GE Healthcare. J.W.G. disclosed no relevant relationships. R.A.B. disclosed no relevant relationships. J.S. disclosed no

relevant relationships. J.E.V. Activities related to the present article: disclosed no

relevant relationships. Activities not related to the present article: is a consultant for

Medicenna !erapeutics. Other relationships: disclosed no relevant relationships.

C.P.H. Activities related to the present article: disclosed no relevant relationships.

Activities not related to the present article: is on the editorial board of Radiology;

has been paid for expert testimony by various medicolegal firms; institution has

grants or grants pending from General Electric; has been paid for meeting and travel

expenses to the Korean Congress of Radiology. Other relationships: disclosed no

relevant relationships. J.K. disclosed no relevant relationships. D.B.V. Activities related to the present article: disclosed no relevant relationships. Activities not related

to the present article: institution has received grants for MRI development from GE

Healthcare. Other relationships: disclosed no relevant relationships.

References

1. Ardenkjaer-Larsen JH, Fridlund B, Gram A, et al. Increase in signal-to-noise

ratio of . 10,000 times in liquid-state NMR. Proc Natl Acad Sci U S A

2003;100(18):10158–10163.

2. Nelson SJ, Kurhanewicz J, Vigneron DB, et al. Metabolic imaging of patients

with prostate cancer using hyperpolarized [1-¹³C]pyruvate. Sci Transl Med

2013;5(198):198ra08.

3. Park JM, Josan S, Grafendorfer T, et al. Measuring mitochondrial metabolism in rat brain in vivo using MR Spectroscopy of hyperpolarized [2-¹³C]

pyruvate. NMR Biomed 2013;26(10):1197–1203.

4. Josan S, Hurd R, Park JM, et al. Dynamic metabolic imaging of hyperpolarized [2-(13) C]pyruvate using spiral chemical shift imaging with alternating

spectral band excitation. Magn Reson Med 2014;71(6):2051–2058.

5. Gallagher FA, Kettunen MI, Hu DE, et al. Production of hyperpolarized

[1,4-13C2]malate from [1,4-13C2]fumarate is a marker of cell necrosis and

treatment response in tumors. Proc Natl Acad Sci U S A 2009;106(47):19801–

19806.

6. Mignion L, Dutta P, Martinez GV, Foroutan P, Gillies RJ, Jordan BF.

Monitoring chemotherapeutic response by hyperpolarized 13C-fumarate

MRS and diffusion MRI. Cancer Res 2014;74(3):686–694.

7. Nielsen PM, Eldirdiri A, Bertelsen LB, Jørgensen HS, Ardenkjaer-Larsen JH,

Laustsen C. Fumarase activity: an in vivo and in vitro biomarker for acute

kidney injury. Sci Rep 2017;7(1):40812.

8. Clatworthy MR, Kettunen MI, Hu DE, et al. Magnetic resonance imaging

with hyperpolarized [1,4-(13)C2]fumarate allows detection of early renal acute

tubular necrosis. Proc Natl Acad Sci U S A 2012;109(33):13374–13379.

9. Miller JJ, Lau AZ, Nielsen PM, et al. Hyperpolarized [1,4-13C2]fumarate

enables magnetic resonance-based imaging of myocardial necrosis. jacc

cardiovasc imaging 2018;11(11):1594–1606.

10. Gallagher FA, Kettunen MI, Day SE, et al. Magnetic resonance imaging

of pH in vivo using hyperpolarized 13C-labelled bicarbonate. Nature

2008;453(7197):940–943.

11. Gallagher FA, Sladen H, Kettunen MI, et al. Carbonic anhydrase activity

monitored in vivo by hyperpolarized 13C-magnetic resonance spectroscopy

demonstrates its importance for pH regulation in tumors. Cancer Res

2015;75(19):4109–4118.

12. Schroeder MA, Swietach P, Atherton HJ, et al. Measuring intracellular pH in

the heart using hyperpolarized carbon dioxide and bicarbonate: a 13C and 31P

magnetic resonance spectroscopy study. Cardiovasc Res 2010;86(1):82–91.

13. Wilson DM, Keshari KR, Larson PE, et al. Multi-compound polarization

by DNP allows simultaneous assessment of multiple enzymatic activities in

vivo. J Magn Reson 2010;205(1):141–147.

14. Keshari KR, Kurhanewicz J, Bok R, Larson PE, Vigneron DB, Wilson DM.

Hyperpolarized 13C dehydroascorbate as an endogenous redox sensor for in

vivo metabolic imaging. Proc Natl Acad Sci U S A 2011;108(46):18606–

18611.

15. Bohndiek SE, Kettunen MI, Hu DE, et al. Hyperpolarized [1-13C]-ascorbic

and dehydroascorbic acid: vitamin C as a probe for imaging redox status in

vivo. J Am Chem Soc 2011;133(30):11795–11801.

eral areas. Importantly, these efforts will require a close collaboration between basic scientists and clinicians, as well as a strong

partnership between academia, industry, and funding agencies.

Commercially available dual-tuned (1

H and 13C) multichannel coils that allow high signal-to-noise ratio imaging for all

body areas are essential. Continuing work to improve pulse sequences that permit the imaging of different organs and with

different HP 13C probes is necessary to increase clinical applicability. Further refinements are also needed to efficiently and

reliably achieve higher and faster polarizations and to simplify

methods of sterile HP probe delivery. To facilitate clinical adoption, automated procedures for HP probe delivery, imaging acquisition, and data analysis are also critical.

As indicated in a recent whitepaper (102), Food and Drug

Administration approval and clinical adoption of HP 13C

MRI will require establishing its clinical utility in subsequent

larger phase II and III trials. Critical to this is the demonstration of reproducibility, which will necessitate the establishment

of uniform standards for probe production, image acquisition

protocol, and data analysis that enable multicenter trials. Early

multicenter phase II trials should validate the findings from single-center trials, and indications supported by small multicenter

trials can then be carried into larger phase II and phase III studies. !ese larger trials should not only evaluate safety and efficacy

end points but also collect information on clinical impact and

cost effectiveness, which are key data for coverage decisions by

the Centers for Medicare and Medicaid Services and other payers. In the case of [1-13C]pyruvate, to assist sites planning clinical

HP studies, the Investigational New Drug (IND) for the use of

[1-13C]pyruvate was transferred to the National Cancer Institute,

and the IND resource is available at https://imaging.cancer.gov/programs_resources/cancer-tracer-synthesis-resources/hyperpolarized-C13-

pyruvate-documentation.htm.

An important consideration in the clinical adoption of HP 13C MRI is cost. !e cost of a clinical polarizer is similar to that

of a PET cyclotron, and the cost of a Good Manufacturing Practice dose of 13C pyruvate is similar to or less than that of many

emerging PET agents. Given the cost, the usage of HP 13C MRI

will likely be restricted to large academic medical centers, as in

the cases of other high-end imaging technologies. Work is ongoing at multiple institutions to further improve the robustness,

reliability, and efficiency of HP 13C MRI (in particular with 13C

pyruvate) as a 5-minute add-on to a standard-of-care multiparametric MRI examination. !is technology has the potential

to provide cost-effective molecular imaging examinations to improve diagnosis and treatment monitoring for various diseases

where current imaging methods show limitations.

Conclusions

HP 13C MRI is an emerging molecular imaging technique that

is actively undergoing clinical translation at multiple institutions. Preclinical results suggest applications in oncology, diabetes, and heart disease, as well as metabolic disease of the liver

and kidney. !e tools required for human studies with HP 13C

MRI are now commercially available, with clinical trials underway. Further expansion into the clinic will require continued

improvement in imaging unit hardware and pulse sequences,

极T代谢磁共振全球科研集锦

51

Wang et al

Radiology: Volume 291: Number 2—May 2019 n radiology.rsna.org 283

as a sensor of temozolomide response in glioblastoma cells. Cancer Res

2014;74(23):7115–7124.

40. Ravoori MK, Singh SP, Lee J, Bankson JA, Kundra V. In vivo assessment

of ovarian tumor response to tyrosine kinase inhibitor pazopanib by using

hyperpolarized 13C-pyruvate MR spectroscopy and 18F-FDG PET/CT imaging in a mouse model. Radiology 2017;285(3):830–838.

41. Saito K, Matsumoto S, Takakusagi Y, et al. 13C-MR spectroscopic imaging

with hyperpolarized [1-13C]pyruvate detects early response to radiotherapy in

SCC tumors and HT-29 tumors. Clin Cancer Res 2015;21(22):5073–5081.

42. Day SE, Kettunen MI, Cherukuri MK, et al. Detecting response of rat

C6 glioma tumors to radiotherapy using hyperpolarized [1- 13C]pyruvate

and 13C magnetic resonance spectroscopic imaging. Magn Reson Med

2011;65(2):557–563.

43. Matsuo M, Kawai T, Kishimoto S, et al. Co-imaging of the tumor oxygenation

and metabolism using electron paramagnetic resonance imaging and 13-C

hyperpolarized magnetic resonance imaging before and after irradiation.

Oncotarget 2018;9(38):25089–25100.

44. Scroggins BT, Matsuo M, White AO, et al. Hyperpolarized [1-13C]-pyruvate

magnetic resonance spectroscopic imaging of prostate cancer In Vivo predicts

efficacy of targeting the Warburg effect. Clin Cancer Res 2018;24(13):3137–

3148.

45. Park JM, Spielman DM, Josan S, et al. Hyperpolarized (13)C-lactate to

(13)C-bicarbonate ratio as a biomarker for monitoring the acute response

of anti-vascular endothelial growth factor (anti-VEGF) treatment. NMR

Biomed 2016;29(5):650–659.

46. Witney TH, Kettunen MI, Hu DE, et al. Detecting treatment response in

a model of human breast adenocarcinoma using hyperpolarised [1-13C]

pyruvate and [1,4-13C2]fumarate. Br J Cancer 2010;103(9):1400–1406.

47. Düwel S, Durst M, Gringeri CV, et al. Multiparametric human hepatocellular

carcinoma characterization and therapy response evaluation by hyperpolarized

(13) C MRSI. NMR Biomed 2016;29(7):952–960.

48. Bohndiek SE, Kettunen MI, Hu DE, Brindle KM. Hyperpolarized (13)C

spectroscopy detects early changes in tumor vasculature and metabolism after

VEGF neutralization. Cancer Res 2012;72(4):854–864.

49. Lee JE, Diederich CJ, Bok R, et al. Assessing high-intensity focused ultrasound

treatment of prostate cancer with hyperpolarized 13 C dual-agent imaging of

metabolism and perfusion. NMR Biomed 2018 Jul 18:e3962 [Epub ahead

of print] https://doi.org/10.1002/nbm.3962.

50. Scholz DJ, Janich MA, Köllisch U, et al. Quantified pH imaging with hyperpolarized (13) C-bicarbonate. Magn Reson Med 2015;73(6):2274–2282.

51. Estrella V, Chen T, Lloyd M, et al. Acidity generated by the tumor microenvironment drives local invasion. Cancer Res 2013;73(5):1524–1535.

52. Lau AZ, Miller JJ, Robson MD, Tyler DJ. Cardiac perfusion imaging using

hyperpolarized (13)C urea using flow sensitizing gradients. Magn Reson Med

2016;75(4):1474–1483.

53. Hagemann CE, Ghotbi AA, Kjær A, Hasbak P. Quantitative myocardial

blood flow with Rubidium-82 PET: a clinical perspective. Am J Nucl Med

Mol Imaging 2015;5(5):457–468.

54. Oh-Ici D, Wespi P, Busch J, et al. Hyperpolarized metabolic MR imaging

of acute myocardial changes and recovery after ischemia-reperfusion in a

small-animal model. Radiology 2016;278(3):742–751.

55. Merritt ME, Harrison C, Storey C, Jeffrey FM, Sherry AD, Malloy CR. Hyperpolarized 13C allows a direct measure of flux through a single enzyme-catalyzed

step by NMR. Proc Natl Acad Sci U S A 2007;104(50):19773–19777.

56. Schroeder MA, Cochlin LE, Heather LC, Clarke K, Radda GK, Tyler

DJ. In vivo assessment of pyruvate dehydrogenase flux in the heart using

hyperpolarized carbon-13 magnetic resonance. Proc Natl Acad Sci U S A

2008;105(33):12051–12056.

57. Atherton HJ, Dodd MS, Heather LC, et al. Role of pyruvate dehydrogenase

inhibition in the development of hypertrophy in the hyperthyroid rat heart:

a combined magnetic resonance imaging and hyperpolarized magnetic

resonance spectroscopy study. Circulation 2011;123(22):2552–2561.

58. Schroeder MA, Lau AZ, Chen AP, et al. Hyperpolarized (13)C magnetic

resonance reveals early- and late-onset changes to in vivo pyruvate metabolism

in the failing heart. Eur J Heart Fail 2013;15(2):130–140.

59. Le Page LM, Rider OJ, Lewis AJ, et al. Increasing pyruvate dehydrogenase flux as a treatment for diabetic cardiomyopathy: a combined 13C

hyperpolarized magnetic resonance and echocardiography study. Diabetes

2015;64(8):2735–2743.

60. Dodd MS, Ball DR, Schroeder MA, et al. In vivo alterations in cardiac metabolism and function in the spontaneously hypertensive rat heart. Cardiovasc

Res 2012;95(1):69–76.

61. Schroeder MA, Atherton HJ, Ball DR, et al. Real-time assessment of Krebs

cycle metabolism using hyperpolarized 13C magnetic resonance spectroscopy.

FASEB J 2009;23(8):2529–2538.

16. Keshari KR, Sai V, Wang ZJ, Vanbrocklin HF, Kurhanewicz J, Wilson DM.

Hyperpolarized [1-13C]dehydroascorbate MR spectroscopy in a murine

model of prostate cancer: comparison with 18F-FDG PET. J Nucl Med

2013;54(6):922–928.

17. Keshari KR, Wilson DM, Sai V, et al. Noninvasive in vivo imaging of diabetesinduced renal oxidative stress and response to therapy using hyperpolarized

13C dehydroascorbate magnetic resonance. Diabetes 2015;64(2):344–352.

18. Wilson DM, Di Gialleonardo V, Wang ZJ, et al. Hyperpolarized 13C spectroscopic evaluation of oxidative stress in a rodent model of steatohepatitis.

Sci Rep 2017;7(1):46014.

19. von Morze C, Larson PE, Hu S, et al. Imaging of blood flow using hyperpolarized [(13)C]urea in preclinical cancer models. J Magn Reson Imaging

2011;33(3):692–697.

20. Fuetterer M, Busch J, Peereboom SM, et al. Hyperpolarized 13C urea

myocardial first-pass perfusion imaging using velocity-selective excitation. J

Cardiovasc Magn Reson 2017;19(1):46.

21. Chen HY, Larson PEZ, Bok RA, et al. Assessing prostate cancer aggressiveness with hyperpolarized dual-agent 3D dynamic imaging of metabolism

and perfusion. Cancer Res 2017;77(12):3207–3216.

22. Lau AZ, Miller JJ, Robson MD, Tyler DJ. Simultaneous assessment of cardiac

metabolism and perfusion using copolarized [1-13 C]pyruvate and 13 C-urea.

Magn Reson Med 2017;77(1):151–158.

23. Albers MJ, Bok R, Chen AP, et al. Hyperpolarized 13C lactate, pyruvate, and

alanine: noninvasive biomarkers for prostate cancer detection and grading.

Cancer Res 2008;68(20):8607–8615.

24. Hu S, Balakrishnan A, Bok RA, et al. 13C-pyruvate imaging reveals alterations

in glycolysis that precede c-Myc-induced tumor formation and regression.

Cell Metab 2011;14(1):131–142.

25. Keshari KR, Sriram R, Koelsch BL, et al. Hyperpolarized 13C-pyruvate

magnetic resonance reveals rapid lactate export in metastatic renal cell carcinomas. Cancer Res 2013;73(2):529–538.

26. Sriram R, Van Criekinge M, DeLos Santos J, et al. Non-invasive differentiation

of benign renal tumors from clear cell renal cell carcinomas using clinically

translatable hyperpolarized 13C pyruvate magnetic resonance. Tomography

2016;2(1):35–42.

27. Golman K, Zandt RI, Lerche M, Pehrson R, Ardenkjaer-Larsen JH. Metabolic imaging by hyperpolarized 13C magnetic resonance imaging for in

vivo tumor diagnosis. Cancer Res 2006;66(22):10855–10860.

28. Harris T, Eliyahu G, Frydman L, Degani H. Kinetics of hyperpolarized

13C1-pyruvate transport and metabolism in living human breast cancer

cells. Proc Natl Acad Sci U S A 2009;106(43):18131–18136.

29. Hu S, Lustig M, Balakrishnan A, et al. 3D compressed sensing for highly

accelerated hyperpolarized (13)C MRSI with in vivo applications to transgenic

mouse models of cancer. Magn Reson Med 2010;63(2):312–321.

30. Kettunen MI, Hu DE, Witney TH, et al. Magnetization transfer measurements of exchange between hyperpolarized [1-13C]pyruvate and [1-13C]

lactate in a murine lymphoma. Magn Reson Med 2010;63(4):872–880.

31. Park I, Larson PE, Zierhut ML, et al. Hyperpolarized 13C magnetic

resonance metabolic imaging: application to brain tumors. Neuro Oncol

2010;12(2):133–144.

32. Park JM, Josan S, Jang T, et al. Metabolite kinetics in C6 rat glioma model

using magnetic resonance spectroscopic imaging of hyperpolarized [1-(13)

C]pyruvate. Magn Reson Med 2012;68(6):1886–1893.

33. Serrao EM, Kettunen MI, Rodrigues TB, et al. MRI with hyperpolarised

[1-13C]pyruvate detects advanced pancreatic preneoplasia prior to invasive

disease in a mouse model. Gut 2016;65(3):465–475.

34. Day SE, Kettunen MI, Gallagher FA, et al. Detecting tumor response to

treatment using hyperpolarized 13C magnetic resonance imaging and spectroscopy. Nat Med 2007;13(11):1382–1387 [Published correction appears

in Nat Med 2007;13(12):1521.] https://doi.org/10.1038/nm1650.

35. Lodi A, Woods SM, Ronen SM. Treatment with the MEK inhibitor U0126

induces decreased hyperpolarized pyruvate to lactate conversion in breast,

but not prostate, cancer cells. NMR Biomed 2013;26(3):299–306.

36. Venkatesh HS, Chaumeil MM, Ward CS, Haas-Kogan DA, James CD, Ronen

SM. Reduced phosphocholine and hyperpolarized lactate provide magnetic

resonance biomarkers of PI3K/Akt/mTOR inhibition in glioblastoma. Neuro

Oncol 2012;14(3):315–325.

37. Chaumeil MM, Ozawa T, Park I, et al. Hyperpolarized 13C MR spectroscopic imaging can be used to monitor Everolimus treatment in vivo in an

orthotopic rodent model of glioblastoma. Neuroimage 2012;59(1):193–201.

38. Park I, Bok R, Ozawa T, et al. Detection of early response to temozolomide

treatment in brain tumors using hyperpolarized 13C MR metabolic imaging.

J Magn Reson Imaging 2011;33(6):1284–1290.

39. Park I, Mukherjee J, Ito M, et al. Changes in pyruvate metabolism detected

by magnetic resonance imaging are linked to DNA damage and serve

极T代谢磁共振全球科研集锦

52

State of the Art and Future Directions of Hyperpolarized Carbon 13 MRI

284 radiology.rsna.org n Radiology: Volume 291: Number 2—May 2019

82. Ardenkjaer-Larsen JH, Leach AM, Clarke N, Urbahn J, Anderson D, Skloss

TW. Dynamic nuclear polarization polarizer for sterile use intent. NMR

Biomed 2011;24(8):927–932.

83. Chen HY, Larson PEZ, Gordon JW, et al. Technique development of 3D

dynamic CS-EPSI for hyperpolarized 13 C pyruvate MR molecular imaging

of human prostate cancer. Magn Reson Med 2018;80(5):2062–2072.

84. Gordon JW, Vigneron DB, Larson PE. Development of a symmetric echo

planar imaging framework for clinical translation of rapid dynamic hyperpolarized 13 C imaging. Magn Reson Med 2017;77(2):826–832.

85. Lau AZ, Chen AP, Ghugre NR, et al. Rapid multislice imaging of hyperpolarized 13C pyruvate and bicarbonate in the heart. Magn Reson Med

2010;64(5):1323–1331.

86. Milshteyn E, von Morze C, Gordon JW, Zhu Z, Larson PEZ, Vigneron

DB. High spatiotemporal resolution bSSFP imaging of hyperpolarized [1-

13 C]pyruvate and [1-13 C]lactate with spectral suppression of alanine and

pyruvate-hydrate. Magn Reson Med 2018;80(3):1048–1060.

87. Tang S, Jiang W, Chen HY, Bok R, Vigneron DB, Larson PEA. A 2DRF

pulse sequence for bolus tracking in hyperpolarized 13C imaging. Magn

Reson Med 2015;74(2):506–512.

88. Golman K, in ’t Zandt R, !aning M. Real-time metabolic imaging. Proc

Natl Acad Sci U S A 2006;103(30):11270–11275.

89. Larson PE, Hu S, Lustig M, et al. Fast dynamic 3D MR spectroscopic imaging

with compressed sensing and multiband excitation pulses for hyperpolarized

13C studies. Magn Reson Med 2011;65(3):610–619.

90. Bankson JA, Walker CM, Ramirez MS, et al. Kinetic modeling and constrained reconstruction of hyperpolarized [1-13C]-pyruvate offers improved

metabolic imaging of tumors. Cancer Res 2015;75(22):4708–4717.

91. Harrison C, Yang C, Jindal A, et al. Comparison of kinetic models for analysis

of pyruvate-to-lactate exchange by hyperpolarized 13 C NMR. NMR Biomed

2012;25(11):1286–1294.

92. Larson PEZ, Chen HY, Gordon JW, et al. Investigation of analysis methods

for hyperpolarized 13C-pyruvate metabolic MRI in prostate cancer patients.

NMR Biomed 2018;31(11):e3997.

93. Aggarwal R, Vigneron DB, Kurhanewicz J. Hyperpolarized 1-[13C]-pyruvate

magnetic resonance imaging detects an early metabolic response to androgen

ablation therapy in prostate cancer. Eur Urol 2017;72(6):1028–1029.

94. Granlund KL, Vargas HA, Lyashchenko SK, et al. Utilizing hyperpolarized

MRI in prostate cancer to assess metabolic dynamics and histopathologic grade

[abstr]. In: Proceedings of the Twenty-Fifth Meeting of the International

Society for Magnetic Resonance in Medicine. Berkeley, Calif: International

Society for Magnetic Resonance in Medicine, 2017; 0727.

95. Park I, Larson PEZ, Gordon JW, et al. Development of methods and feasibility of using hyperpolarized carbon-13 imaging data for evaluating brain

metabolism in patient studies. Magn Reson Med 2018;80(3):864–873.

96. Miloushev VZ, Granlund KL, Boltyanskiy R, et al. Metabolic imaging of

the human brain with hyperpolarized 13C pyruvate demonstrates 13C lactate

production in brain tumor patients. Cancer Res 2018;78(14):3755–3760.

97. Cunningham CH, Lau JY, Chen AP, et al. Hyperpolarized 13C metabolic MRI

of the human heart: initial experience. Circ Res 2016;119(11):1177–1182.

98. Lau AZ, Chen AP, Lau JY, et al. Hyperpolarized 13C metabolic imaging

of human hypertrophic cardiomyopathy [abstr]. In: Proceedings of the

Twenty-Sixth Meeting of the International Society for Magnetic Resonance

in Medicine. Berkeley, Calif: International Society for Magnetic Resonance

in Medicine, 2018; 0276.

99. Salminen E, Hogg A, Binns D, Frydenberg M, Hicks R. Investigations with

FDG-PET scanning in prostate cancer show limited value for clinical practice.

Acta Oncol 2002;41(5):425–429.

100.Keshari KR, Sriram R, Van Criekinge M, et al. Metabolic reprogramming and validation of hyperpolarized 13C lactate as a prostate cancer

biomarker using a human prostate tissue slice culture bioreactor. Prostate

2013;73(11):1171–1181.

101.Izuishi K, Yamamoto Y, Mori H, et al. Molecular mechanisms of [18F]fluorodeoxyglucose accumulation in liver cancer. Oncol Rep 2014;31(2):701–706.

102.Kurhanewicz J, Vigneron DB, Ardenkjaer-Larsen JH, et al. Hyperpolarized

13C MRI: path to clinical translation in oncology. Neoplasia 2019;21(1):1–16.

62. Dodd MS, Atherton HJ, Carr CA, et al. Impaired in vivo mitochondrial Krebs

cycle activity after myocardial infarction assessed using hyperpolarized magnetic

resonance spectroscopy. Circ Cardiovasc Imaging 2014;7(6):895–904.

63. Merritt ME, Harrison C, Sherry AD, Malloy CR, Burgess SC. Flux through

hepatic pyruvate carboxylase and phosphoenolpyruvate carboxykinase

detected by hyperpolarized 13C magnetic resonance. Proc Natl Acad Sci

U S A 2011;108(47):19084–19089.

64. Moon CM, Oh CH, Ahn KY, et al. Metabolic biomarkers for non-alcoholic

fatty liver disease induced by high-fat diet: in vivo magnetic resonance spectroscopyofhyperpolarized [1-13C] pyruvate. Biochem Biophys Res Commun

2017;482(1):112–119.

65. Josan S, Billingsley K, Orduna J, et al. Assessing inflammatory liver injury in

an acute CCl4 model using dynamic 3D metabolic imaging of hyperpolarized

[1-(13)C]pyruvate. NMR Biomed 2015;28(12):1671–1677.

66. Moon CM, Shin SS, Lim NY, et al. Metabolic alterations in a rat model of

hepatic ischaemia reperfusion injury: in vivo hyperpolarized 13 C MRS and

metabolic imaging. Liver Int 2018;38(6):1117–1127.

67. Morze CV, Allu PKR, Chang GY, et al. Non-invasive detection of divergent

metabolic signals in insulin deficiency vs. insulin resistance in vivo. Sci Rep

2018;8(1):2088.

68. Marco-Rius I, von Morze C, Sriram R, et al. Monitoring acute metabolic

changes in the liver and kidneys induced by fructose and glucose using hyperpolarized [2-13 C]dihydroxyacetone. Magn Reson Med 2017;77(1):65–73.

69. Moreno KX, Satapati S, DeBerardinis RJ, Burgess SC, Malloy CR,

Merritt ME. Real-time detection of hepatic gluconeogenic and glycogenolytic states using hyperpolarized [2-13C]dihydroxyacetone. J Biol Chem

2014;289(52):35859–35867.

70. Ohliger MA, von Morze C, Marco-Rius I, et al. Combining hyperpolarized

13 C MRI with a liver-specific gadolinium contrast agent for selective assessment of hepatocyte metabolism. Magn Reson Med 2017;77(6):2356–2363.

71. Laustsen C, Østergaard JA, Lauritzen MH, et al. Assessment of early diabetic

renal changes with hyperpolarized [1-(13) C]pyruvate. Diabetes Metab Res

Rev 2013;29(2):125–129.

72. Laustsen C, Lycke S, Palm F, et al. High altitude may alter oxygen availability

and renal metabolism in diabetics as measured by hyperpolarized [1-(13)C]

pyruvate magnetic resonance imaging. Kidney Int 2014;86(1):67–74.

73. Baligand C, Qin H, True-Yasaki A, et al. Hyperpolarized 13 C magnetic

resonance evaluation of renal ischemia reperfusion injury in a murine model.

NMR Biomed 2017;30(10):e3765.

74. Nielsen PM, Laustsen C, Bertelsen LB, et al. In situ lactate dehydrogenase

activity: a novel renal cortical imaging biomarker of tubular injury? Am J

Physiol Renal Physiol 2017;312(3):F465–F473.

75. Nielsen PM, Szocska Hansen ES, Nørlinger TS, et al. Renal ischemia and

reperfusion assessment with three-dimensional hyperpolarized 13 C,15 N2-

urea. Magn Reson Med 2016;76(5):1524–1530.

76. Bertelsen LB, Nielsen PM, Qi H, et al. Diabetes induced renal urea transport

alterations assessed with 3D hyperpolarized 13 C,15 N-Urea. Magn Reson

Med 2017;77(4):1650–1655.

77. !ind K, Jensen MD, Hegarty E, et al. Mapping metabolic changes associated

with early radiation induced lung injury post conformal radiotherapy using

hyperpolarized ¹³C-pyruvate magnetic resonance spectroscopic imaging.

Radiother Oncol 2014;110(2):317–322.

78. Guglielmetti C, Najac C, Didonna A, Van der Linden A, Ronen SM,

Chaumeil MM. Hyperpolarized 13C MR metabolic imaging can detect

neuroinflammation in vivo in a multiple sclerosis murine model. Proc Natl

Acad Sci U S A 2017;114(33):E6982–E6991.

79. MacKenzie JD, Yen YF, Mayer D, Tropp JS, Hurd RE, Spielman DM.

Detection of inflammatory arthritis by using hyperpolarized 13C-pyruvate

with MR imaging and spectroscopy. Radiology 2011;259(2):414–420.

80. Sriram R, Nguyen J, Santos JD, et al. Molecular detection of inflammation in cell models using hyperpolarized 13C-pyruvate. !eranostics

2018;8(12):3400–3407.

81. Najac C, Chaumeil MM, Kohanbash G, et al. Detection of inflammatory

cell function using (13)C magnetic resonance spectroscopy of hyperpolarized

[6-(13)C]-arginine. Sci Rep 2016;6(1):31397.

极T代谢磁共振全球科研集锦

54

极T代谢磁共振全球科研集锦

55

肿瘤篇

极T代谢磁共振全球科研集锦

极T代谢磁共振全球科研集锦

56

Development of Methods and Feasibility of

Using Hyperpolarized Carbon-13 Imaging

Data for Evaluating Brain Metabolism in Patient Studies

研究背景

研究结果

研究对象

研究过程

ג极ࣅ༐ 13 (13C) 代谢ׯ๟ၟᅃዖᆩᇀೠࠚ้ํ代谢ڦ࿮ׯظၟݛ๕ăԨ研৯ڦణڦ๟ਸ݀ࢅํแ๑ᆩ [1-13C] եཛྷ

໗༑֪స዗ୀࢅ໱ഄหঢ়ဣཥदթ࣒ኁ༹ా代谢ڦํᄓ֧୼ă

ٗ 2 ఁথ๴ዎଐܠڦႚႠঌዊజဦԇୀ࣒ኁڦ 2 ૑௝ೝ௬ݴՐ

ాࢅ೪঴ڦऺࠚ਍ༀຕۯ EPSI 2D ڦइൽ܈ࢩೌൎ௝૑ 2 ࢅ୲

֭ג极ࣅ 13C 代谢࿿Բ୲཮ၟ۠ेă ֒෥۠े཮ၟڦԲ૩ֶᅴ

ࡃᅺᇀᆩᇀ࣒ኁ 2 ڇڦհ࢛܎ۨ݋ገঙघ݀ݛӄࢅᆩᇀ࣒ኁ 8

ܠڦհ܎੗Վ݋ገঙघ݀ݛӄă൩ጀᅪLjሞኄଇዖ൧઄ူLj༐

໗ൠჸ / եཛྷ໗ჸۼ ๟ሞ NAB ں߸൶ߛă ฉݛၟ཮ዐ࣒ڦኁ

DŽ࣒ኁ 2DžሞԨْॠֱ้࣒ᆶ৊ႜႠ዗ୀLjူݛၟ཮ዐ࣒ڦኁ

DŽ࣒ኁ 8Dž኱ڟໜڦࢫໜݡ෢௮֍ਏᆶ৊ቛڦ༬ኙă lac/pyr ࢅ

bicarb/pyr ݴ՚代՗ළ໗ჸ / եཛྷ໗ჸࢅ༐໗ൠჸ / եཛྷ໗ჸă

8ఁ࣒ኁǖసୀ

ג极ࣅ༐ 13 (13C) 磁共振 (MR) 代谢ׯ๟ၟᅃዖۉݥ૗ĂݣݥพႠׯၟݛ݆Lj੗ᆩᇀ֪ଉํ้代谢ăම঴ۯༀࢃ极ࣅ

(DNP) ڦፌႎ݀ቛྺೠࠚా༹代谢༵ࠃକᅃዖସට႗ݛڦݺ݆Ljᇑدཥڦ 13C MR ݛ݆၎ԲLj႑ࡽഽ܈ᆶକ਎ڦۇٴ༵

ߛLjժ൐ీࠕሞ܌้क़ాइڥਏᆶߛ੣क़ݴՐ୲ڦ 13C 代谢ׯၟຕ਍ă

ኄᇑ֪॔዗ୀิࢅ׊ೠܔࠚዎଐݒڦᆌᆮഄ၎࠲LjᅺྺܱႠဦԇঢ়׉ᆶฉۙڦළ໗ྃൠா ALj๟ۙব޿཰০ڦாă

Ԩ研৯ڦణڦ๟૧ᆩටసዐڦ] 1-13C] եཛྷ໗ਸ݀ࢅํแג极ࣅ 13C 代谢ׯၟLjഄ׊೺ణՔ๟ీࠕ֪॔స዗ୀࢅ໱ഄ

หঢ়ဣཥदթ࣒ኁڦా༹代谢ăྺُᆌᆩยऺڦ 13C ၍ංࢅஞ؋Ⴞଚ๯ံሞ༹ఇዐ৊ႜକ֪๬ă඗ࢫံٗമኑ܏ྺ

ঌዊୀDŽׯටዐፌ׉९ڦᇱ݀Ⴀస዗ୀDž࣒ڦኁዐइۯڥༀ 13C ຕ਍ă

极T代谢磁共振全球科研集锦

57

结 论

应用方向

ሞටసዐํแג极ࣅ 13C 代谢ׯၟڦํᄓ֧୼ᅙԥਸ݀Lj؛օ࣒ڦኁ研৯ኤํକ๑ᆩ޿रຍڦҾ全Ⴀࢅ੗ႜႠă इ

ڦڥ঳ࡕ௽՗Ljג极ڦࣅ] 1-13C] եཛྷ໗ԥገሏࡗحეసೡቱLjժሞ੗๑ᆩԨ研৯ยऺڦஞ؋Ⴞଚࢅพೕ၍ං֪ଉڦ

้क़ݔాྷገࣅྺ代谢ׂ࿿ [1-13C] ළ໗ࢅ 13C ༐໗ൠჸă৑࣏࠶Ⴔᄲ৊ᅃօᆫࣅLjڍኄၵ؛օ݀၄ኧܔ׼Ԩरຍሞ

స዗ୀࢅ໱ഄหঢ়ဣཥदթ࣒ኁዐڦ৊ᅃօ研৯ă

స዗ୀ

极T代谢磁共振全球科研集锦

58

FULL PAPER

Development of Methods and Feasibility of Using

Hyperpolarized Carbon-13 Imaging Data for Evaluating

Brain Metabolism in Patient Studies

Ilwoo Park,1 Peder E.Z. Larson ,

2 Jeremy W. Gordon,2 Lucas Carvajal,2

Hsin-Yu Chen,2 Robert Bok,2 Mark Van Criekinge,2 Marcus Ferrone,3

James B. Slater,2 Duan Xu,2 John Kurhanewicz,2 Daniel B. Vigneron,2,4

Susan Chang,5 and Sarah J. Nelson2,4*

Purpose: Hyperpolarized carbon-13 (13C) metabolic imaging is

a noninvasive imaging modality for evaluating real-time metabolism. The purpose of this study was to develop and implement experimental strategies for using [1-13C]pyruvate to

probe in vivo metabolism for patients with brain tumors and

other neurological diseases.

Methods: The 13C radiofrequency coils and pulse sequences

were tested in a phantom and were performed using a 3 Tesla

whole-body scanner. Samples of [1-13C]pyruvate were polarized using a SPINlab system. Dynamic 13C data were acquired

from 8 patients previously diagnosed with brain tumors, who

had received treatment and were being followed with serial

magnetic resonance scans.

Results: The phantom studies produced good-quality spectra

with a reduction in signal intensity in the center attributed to

the reception profiles of the 13C receive coils. Dynamic data

obtained from a 3-cm slice through a patient’s brain following

injection with [1-13C]pyruvate showed the anticipated arrival

of the agent, its conversion to lactate and bicarbonate, and

subsequent reduction in signal intensity. A similar temporal

pattern was observed in 2D dynamic patient studies, with

signals corresponding to pyruvate, lactate, and bicarbonate

being in normal appearing brain, but only pyruvate and lactate being detected in regions corresponding to the anatomical lesion. Physiological monitoring and follow-up confirmed

that there were no adverse events associated with the

injection.

Conclusion: This study has presented the first application of

hyperpolarized 13C metabolic imaging in patients with brain

tumor and demonstrated the safety and feasibility of using

hyperpolarized [1-13C]pyruvate to evaluate in vivo brain

metabolism. Magn Reson Med 000:000–000, 2018. VC 2018

International Society for Magnetic Resonance in Medicine.

Key words: brain tumor patients; dynamic nuclear polarization; hyperpolarized carbon-13 MRI

INTRODUCTION

Hyperpolarized carbon-13 (13C) magnetic resonance (MR)

metabolic imaging is a nonionizing, nonradioactive imaging method that can be used to measure real-time metabolism. The recent development of dissolution dynamic

nuclear polarization (DNP) offers an exciting method for

assessing in vivo metabolism, with a huge gain in signal

intensity over the conventional 13C MR methods (1) and

enables the acquisition of 13C metabolic imaging data

with high spatial resolution in a short time (2). One of

the first applications of this technology has been to evaluate the conversion of [1-13C]pyruvate to [1-13C]lactate.

This is particularly relevant for monitoring tumor growth

and assessing response to therapy because malignant

cells frequently have upregulated lactate dehydrogenase

A, which is the enzyme that regulates this pathway (3,4).

Preclinical studies that have shown promising results

include the assessment of a wide range of different cancers, as well as cardiac disease, traumatic brain injury,

and multiple sclerosis (5–13).

The first-in-human study using hyperpolarized 13C

metabolic imaging was performed in patients with prostate cancer and was able to demonstrate the safety and

feasibility of the technology in a clinical setting (14).

This has led to great interest in applying similar methods

to other groups of patients and also for volunteer cardiac

studies (15). The purpose of the current study was to

develop and implement hyperpolarized 13C metabolic

imaging using [1-13C]pyruvate in the human brain, with

the long-term goal of being able to monitor in vivo

metabolism for patients with brain tumors and other

neurological diseases. The 13C coils and pulse sequences

designed for this application were first tested in phantoms. Dynamic 13C data were then obtained from

patients with a prior diagnosis of glioma, which is the

most common primary brain tumor in adults.

1

Department of Radiology, Chonnam National University Medical School

and Hospital, Gwangju, Korea.

2

Department of Radiology and Biomedical Imaging, University of California,

San Francisco, California, USA.

3

Department of Clinical Pharmacy, University of California, San Francisco,

California, USA.

4

Department of Bioengineering and Therapeutic Sciences, University of

California, San Francisco, California, USA.

5

Department of Neurological Surgery, University of California, San Francisco, California, USA.

Grant sponsor: NIH; Grant numbers: P41EB013598, R21CA170148,

P01CA118816 and R01EB017449.

*Correspondence to: Sarah J. Nelson, Ph.D., University of California, 1700

4th Street, Box 2532, BH-303, San Francisco, CA 94158, USA.

E-mail: Sarah.Nelson@ucsf.edu

Received 13 August 2017; revised 27 November 2017; accepted 16

December 2017

DOI 10.1002/mrm.27077

Published online 00 Month 2018 in Wiley Online Library (wileyonlinelibrary.

com).

Magnetic Resonance in Medicine 00:00–00 (2018)

VC 2018 International Society for Magnetic Resonance in Medicine 1

极T代谢磁共振全球科研集锦

59

METHODS

13C MR setup

All experiments were performed using a 3 Tesla (T) clinical magnetic resonance imaging (MRI) system (GE

Healthcare, Waukesha, WI) with 50-mT/m, 200-mT/m/

ms gradients and a multinuclear spectroscopy hardware

package. The 13C radiofrequency (RF) coil configuration

comprised a bore-insertable 13C volume coil for transmission (16) and a bilateral eight-channel phased-array coil

for reception (17) (Fig. 1a,b). The standard patient headrest holder was modified so that the subjects were comfortably stabilized in the 13C RF coil setup, whereas the

two phased-array receive coils were placed around the

head (Fig. 1c).

13C coil loading tests

To investigate the effect of coil loading on the 13C signal,

a B1 þ map was acquired using the double-angle method

(18) in both unloaded and loaded scenarios using a

head-shaped phantom containing unenriched ethylene

glycol (HOCH2CH2OH, anhydrous, 99.8%; SigmaAldrich, St. Louis, MO). To acquire the B1 þ map, a

single-band spectral-spatial RF pulse (full width at half

maximum ¼ 120 Hz) was used to selectively excite the

central 13C resonance of ethylene glycol and avoid artifacts arising from chemical shift, and then encoded with

a single-shot symmetric echo-planar readout (19). The

total readout duration was 15.46 ms (24 echoes, 0.644

ms echo spacing), with a ¼ 30 #, 2a ¼ 60 #, and 100

averages (7.5 $ 7.5 mm in-plane resolution, 24 $ 24 cm

field of view [FOV], 32 $ 32 matrix, 10-cm slice thickness, and repetition time [TR]/echo time [TE] ¼ 3,000/

15.6 ms; the total scan time was 10 minutes). The B1

map was acquired unloaded (with just the head phantom) and loaded (with the head phantom plus saline).

Loading was confirmed with a change from –7 to –11 dB

using a network analyzer. The flip angles over the ethylene glycol phantom were assessed as grayscale maps and

histograms and compared between the loaded and

unloaded cases.

Phantom tests

The phantom was scanned using the clamshell/phased

array coil configuration shown in Figure 1. For these

scans, the distance between the center of the two receive

coils was 17 cm. 13C spectral data were acquired from a

2-cm slice using a dynamic 13C 2D echo-planar spectroscopic imaging (EPSI) sequence with TR/TE ¼ 3,000/6.1

ms, 20 $ 20 mm2 nominal in-plane resolution (20), 10

phase encodes in the right-left (RL) direction, and a symmetric echo-planar readout in the anteroposterior (AP)

direction with a constant 90 # flip angle excitation.

Patient population

Eight patients who had a prior diagnosis of glioma were

recruited from the neuro-oncology clinic at our institution. Table 1 summarizes their clinical parameters. They

had all received multiple treatments and were being

FIG. 1. 13C RF coil configurations developed for human brain study. (a) Eight-channel 13C phased-array coils. (b) Clamshell volumetric 13C transmit coil and bilateral eight-channel phased-array receive coils. (c) A picture of 13C RF coil setup with a volunteer.

Table 1

Patient Characteristics and QC Values for the 13C Injections

Patient

no. Sex

Weight

(kg)

Initial

diagnosis

age

(years)

Time to

scan

(years)a

Current

diagnosis

Subsequent

clinical

status

EPA

conc

(mM)

pyr

conc

(nM) % polariz-ation pH

Time to

injection

(sec)

1 M 81.4 35 19 OD2 Stable 0.50 247 33 7.3 81

2 M 73.5 30 8 GBM Progressed 1.40 236 38 7.5 84

3 M 75.8 38 6 OD2 Reop/AA 0.30 223 39 8.1 114

4 M 93.0 25 12 AA Reop/AA 3.00 221 43 7.8 86

5 F 81.2 58 2 GBM Progressed 1.30 249 43 7.0 96

6 M 99.4 36 12 OD3 Stable 0.90 239 32 7.4 85

7 F 65.0 46 2 GBM Reop/TE 0.20 232 33 7.9 74

8 F 78.0 54 1 GBM Stable 0.30 242 35 7.2 87

a

Time from initial diagnosis to the 13C scan.

OD2, oligodenroglioma grade 2; OD3, oligodendroglioma grade 3; AA, anaplastic astrocytoma (grade 3); GBM, glioblastoma (grade 4);

TE, treatment effect; EPA, electron paramagnetic agent.

2 Park et al.

极T代谢磁共振全球科研集锦

60

followed with MRI. An Investigational New Drug (IND)

had been obtained from the U.S. Food and Drug Administration for generating the agent and implementing the

clinical protocol. Patients provided written informed

consent for participation in the study, which had institutional review board approval. Electrocardiogram monitoring was performed at baseline and within 1 hour after

the pyruvate injection. Clinical follow-up assessments

were performed at 24 hours.

Sample formulation and polarization

For patients scans, the hyperpolarized [1-13C]pyruvate

was produced using a SPINlab (General Electric, Niskayuna, NY) DNP polarizer that is adjacent to the 3T MR

scanner. The pharmacy kit (fluid path) was filled in an

ISO 5 environment utilizing an isolator (Getinge Group,

Getinge, France) and a clean bench laminar flow hood.

The mixture used for polarization comprised 1.432 g of

[1-13C]pyruvic acid (MilliporeSigma, Miamisburg, OH)

and 28 mg of trityl radical (GE Healthcare, Oslo, Norway). The fluid path was then loaded into one of the

available channels in the SPINlab polarizer. After

approximately 2.5 hours of microwave irradiation at 140

GHz, the mixture of [1-13C]pyruvic acid and trityl radical

was dissolved in sterile water and forced through a filter

that removed trityl radical to a level below 3mM. The

solution was then collected in a receiver vessel, neutralized, and diluted with a sodium hydroxide tris(hydroxymethyl)aminomethane/ethylenediaminetetraacetic acid

buffer solution. The receive assembly that accommodates

quality-control (QC) processes provided rapid measurements of pH, temperature, residual electron paramagnetic agent (EPA) concentration, volume, pyruvate

concentration, and polarization level. The final step in

the automated compounding procedure was for the drug

product to be passed through a sterilizing filter (0.2mm;

ZenPure, Manassas, VA) within the SPINlab QC system

immediately before being collected in a sterile Medrad

syringe.

Once the preparation was complete, a sterile filter

integrity test was performed in parallel with the pyruvate

solution being transported to the MRI scan room and

being set up on a power injector. Successful QC and

filter integrity tests were required before the hyperpolarized pyruvate doses were released for patient injections.

The acceptance criteria for the hyperpolarized pyruvate

injection were: 1) polarization > ¼15%; 2) pyruvate concentration between 220 and 280 mM; 3) EPA concentration < ¼3.0mM; 4) pH between 5.0 and 9.0; 5)

temperature between 25.0 and 37.0 "

C; 6) volume > 38 mL; and 7) filter integrity passes the bubble

point test at 50 psi. Following approval from the pharmacist, a sample corresponding to a dose of 0.43 mL/kg

from the approximately 250-mM pyruvate solution was

delivered to the subject at a rate of 5 mL/s. After completion of pyruvate injection, a further 20 mL of sterile

saline was injected at a rate of 5 mL/s. The polarization,

QC, and timing parameters are summarized in Table 1.

Imaging protocol for patient study

Before the start of each examination, patients were monitored to establish their baseline vital signs and an intravenous catheter was placed in their antecubital vein.

After positioning in the scanner with the 13C coil setup

covering as much of the lesion as possible, T2-weighted

fast spin echo images (TR/TE ¼ 60/4,000 ms, 26-cm FOV,

192 # 256 matrix, 5-mm slice thickness, and number of

excitations ¼ 2) were acquired with the 1

H body coil to

provide an anatomical reference. Frequency calibration

was performed with the 13C coils using the sealed standard that is housed within one of the eight-channel

phased array elements and contains 1 mL of 8 M of 13Curea. Once the appropriate scan parameters had been

defined, the operators of the SPINlab system started the

dissolution process. When the pharmacist had approved

the results provided by the QC system and filter integrity

tests, the pyruvate solution and saline flush were administered into the patient. Dynamic 13C data were acquired

starting 5 seconds after the end of the saline injection

from an axial slab that was centered over the anatomical

lesion.

The acquisition parameters for the 13C data are summarized in Table 2. For patient 1, the dynamic data were

acquired from a 30-mm axial slice with a 10 " flip angle,

TR/TE ¼ 3,000/35 ms, 3-second temporal resolution, and

40 total time points. For 5 patients, 2D-localized

Table 2

Summary of the Acquisition Parameters for the Patient 13C Data, as Well as the Time Point and SNR for Maximum Pyruvate and Lactate

Patient no. Scan typea

Flip

angle ("

)

No. of time

points

Voxel size

(RL # AP # SI cm)

Matrix

sizeb

Peak time

point (s) Delay

Maximum

SNR

pyr lac Dt (s) pyr lac

1 Slice-select 10 40 3 cm slice — 15 24 9 3,266 293

2 2D EPSI 10 24 2 # 2 # 2 10 # 18 6 15 9 754 103

3 2D EPSI 10 24 2 # 2 # 2 10 # 18 9 18 9 286 38

4 2D EPSI 10 16 2 # 2 # 2 10 # 18 3 12 9 665 53

5 2D EPSI 10 24 1.8 # 1.8 # 2 10 # 18 3 12 9 714 38

6 2D EPSI 10 24 1.5 # 1.5 # 2 12 # 18 3 12 9 388 25

7 2D EPSI-mb Variable 10 2 # 2 # 3 10 # 18 6 12 6 119 83

8 2D EPSI-mb Variable 10 2 # 2 # 2 10 # 18 3 12 9 169 91

All studies were acquired with a temporal resolution of 3 seconds. lac and pyr represent lactate and pyruvate, respectively.

a

Slice-select, 2D EPSI, and 2D EPSI-mb represent slice-selective dynamic, 2D-localized dynamic echo-planar spectroscopic imaging,

and multiband RF 2D-localized dynamic echo-planar spectroscopic imaging pulse sequence, respectively.

b

Matrix size in phase encode (RL) # EPSI (AP) directions.

Hyperpolarized Carbon-13 Metabolic Imaging of Patients With Brain Tumors 3

极T代谢磁共振全球科研集锦

61

dynamic EPSI data were obtained from a 20-mm axial

slice with a constant 10 ! flip angle, 24 total time points,

either 10 or 12 phase encodes in the RL direction, and a

symmetric echo-planar readout in the AP direction (20).

The TR/TE was 130/6.1 ms, the time resolution was 3

seconds, and the nominal in-plane spatial resolution was

either 15 " 15, 18 " 18, or 20 " 20 mm2 (21). The EPSI

readout contained the following parameters: spectral resolution ¼ 10.4 Hz, EPSI duration ¼ 96.4 ms, spectral

bandwidth ¼ 543 Hz, minimum resolution ¼ 4.8 mm,

signal-to-noise ratio (SNR) efficiency ¼ 0.94. Similar

dynamic EPSI data were obtained for the other 2

patients, but with an excitation scheme that utilized a

multiband RF pulse and progressively increasing flip

angle scheme (22,23). These flip angle schemes were

designed to efficiently use the magnetization in several

ways: use lower flip angles on pyruvate to increase the

magnetization available for metabolic conversion, evenly

distribute magnetization throughout the acquisition by

accounting for T1 decay, prior RF excitations, and metabolic conversion; and use all available magnetization by

the end of the experiment (23). The plot of pyruvate, lactate, and bicarbonate flip angles across all excitations are

shown in Supporting Figure S1. At the completion of the

13C examination, patients were taken out of the scanner

for postinjection monitoring and then brought back for a

subsequent standard 1

H MR examination that was

obtained with a conventional head coil.

Data analysis

The slice-localized dynamic 13C data were processed

with MATLAB (version 7.0; The MathWorks, Inc.,

Natick, MA). Individual free induction decays were apodized with a 10-Hz Gaussian filter in the time domain

and Fourier-transformed to produce 13C spectra at each

time point. The 2D EPSI dynamic 13C data were processed with software developed in our laboratory including the Spectroscopic Image Visualization and

Computing (SIVIC) package (24,25). The odd and even

echoes of the symmetric EPSI readout were separated

and processed by the following steps: 1) apodization by

a 10-Hz Gaussian filter in the time domain; 2) correction

of timing delays between spatial k-space samples; 3)

ramp samples were gridding onto a uniform grid; 4)

Fourier-transformed; and 5) combined with linear phase

added to correct for timing differences between the even

and odd data. This produced a time series of spectroscopic imaging data for each coil element. In order to

correct for the aliased bicarbonate chemical shift, the

second set of data were generated by demodulating the

raw data at a different frequency before reconstruction.

The coil combination algorithm used for the slice select

and initial review of the 2D EPSI data utilized a magnitude sum of squares algorithm. Subsequent quantitative

analysis of the 2D EPSI data utilized a linear combination of phase-sensitive spectra from different receive

coils with phases and weights determined from the

intensity of the pyruvate signal from the time point at

which it was at a maximum. In addition to the dynamic

analysis, a single spectral array was calculated by summing the time series on a voxel-by-voxel basis. Peak

intensities for lactate and pyruvate were estimated from

the spectral array that was reconstructed at the original

reference frequency, and the peak intensities for bicarbonate were estimated from the array that was reconstructed at its frequency. To perform this analysis, each

array was baseline subtracted, then frequency and phase

corrected using methods described previously for H-1

data (25). SNRs were calculated as the peak height over

the standard deviation from regions in the spectra without any metabolic signal.

To relate the 13C data with anatomical features, the

fluid attenuation inversion recovery (FLAIR) and T1-

weighted post-Gd (gadolinium) images obtained from the

subsequent 1

H imaging examination were aligned with

the body-coil T2-weighted images. A mask corresponding

to brain parenchyma was obtained using the FSL brain

extraction tool. The T2 lesion was defined using manual

segmentation. This was then subtracted from the mask of

brain parenchyma in order to unambiguously define

regions of normal-appearing brain (NAB) tissue. The following criteria were used to classify the voxels for NAB

or T2 lesion: 1) Voxels from NAB were defined as being

at least 80% within the brain, having no overlap with

the T2 lesion, and having lactate SNR greater than 10.0

and bicarbonate SNR greater than 5.0; 2) for the T2

lesion, the voxels that were overlapped by more than

30% with the T2 lesion and had lactate SNR greater than

10.0 were chosen.

RESULTS

13C coil loading test

Figure 2A shows the results from the 13C coil loading

test. The B1 þ field measured using the double-flip-angle

method (26) appeared to be similar between the

unloaded and loaded conditions. The calculated flip

angles from the flip angle maps and histograms were

28 ! 6 7 ! (n ¼ 3) and 26 ! 6 8 ! (n ¼ 3) for the unloaded

and loaded condition, respectively. This meant that the

power requirements did not change between the loaded

and unloaded conditions, and a power calibration on the

head phantom could be used for patient studies. Based

on these results, the ethylene glycol phantom shown was

used to calibrate the transmit gain for the 13C sequence

before subjects were placed on the scanner.

Phantom data

The 13C coils and sequences detected 13C signal from

the phantom (Fig. 2b), but with 3- to 4-fold lower intensity in the center of the FOV attributed to the reception

profiles of the receive coil elements. This is in agreement with previous tests done in a study that focused

on evaluating data from the brain of a nonhuman primate (21).

Patient data

All patients tolerated the pyruvate injection well, and no

adverse effects were observed or reported subsequently.

Levels of polarization and QC parameters obtained are

presented in Table 1 and were well within the specifications defined in the IND. The mean EPA concentration

4 Park et al.

极T代谢磁共振全球科研集锦

62

was 0.99mM (range, 0.2–3.0), pyruvate concentration

was 236 mM (range, 221–249), polarization was 37%

(range, 32–43), and pH was 7.5 (range, 7.0–8.1). The

mean scan start time measured from the start of dissolution was 88 seconds (range, 74–114).

Figure 3 shows the slice-localized 13C dynamic data

from a 30-mm-thick axial slice for subject 1. The location

of the slice is defined on the sagittal image in Figure 3A.

The 13C dynamic spectra are represented as a stack plot

in Figure 3B. They are displayed in magnitude mode

and were acquired with a time resolution of 3 seconds.

SNRs of pyruvate, lactate, and bicarbonate are plotted as

a function of time in Figure 3c. The pyruvate signal at

173 parts per million (ppm) reached a maximum with

SNR of 3,266 at approximately 15 seconds from the start

of the data acquisition, whereas the lactate signal at

185 ppm reached a maximum SNR of 293 at 24 seconds,

which was 9 seconds later. The pyruvate signal

decreased rapidly from the maximum peak. Pyruvatehydrate was observed as a peak lying between the pyruvate and lactate resonances, but had relatively low signal

amplitude (Fig. 3b). Signal from bicarbonate was

detected at approximately 162 ppm and had lower intensity than lactate (Fig. 3c).

The peak time point and maximum voxel SNR for the

2D EPSI dynamic data are shown in Table 2. Although

the time at which the pyruvate delivery was a maximum

varied from 3 to 9 seconds, the additional delay for

reaching the highest lactate was 9 seconds for 6 of the 7

patients and 6 seconds for 1 patient. As expected, the

maximum SNRs was lowest for patient 3, for whom the

time to injection was the longest at 114 seconds, and the

second lowest for patient 6, for whom the number of

phase encodes was increased and the voxel resolution

reduced. The maximum SNR for pyruvate was lower in

patients 7 and 8 because of the multiband excitation and

flip angle scheme that was used, but this scheme

resulted in higher levels of lactate and bicarbonate.

Taken as a whole, these results confirmed that hyperpolarized pyruvate was able to cross the blood–brain barrier and was converted to lactate with a similar time

course between subjects.

The acquisition of the 2D EPSI dynamic 13C metabolic

imaging data made it possible to investigate spatial variations in metabolism. Figure 4 shows reference anatomical images and spectra from multiple time points for

voxels in the T2 lesion (black box) and voxels from contralateral brain tissue (white box) for patient 2. The

FIG. 2. Results from the 13C coil loading and initial data acquisition tests. (a) B1 þ maps of flip angle were calculated using a double-flipangle method and showed that the B1 maps were similar between the unloaded and loaded conditions. (b) 2D EPSI data from the

head-shaped phantom containing ethylene glycol demonstrated the combined reception profile of the receive coils.

FIG. 3. Slice-localized hyperpolarized 13C dynamic data from patient 1. (a) T2 FLAIR image in a sagittal plane shows the hyperintense

region around the resection cavity and the ventricle. (b) The stack plot of 13C magnitude spectra, showing a temporal evolution of lactate, pyruvate, and bicarbonate signal from the brain. (c) The SNR of lactate, pyruvate, urea, and bicarbonate are plotted over time. The

pyruvate SNR was divided by 4 so that it could be viewed on the same graph as the pyruvate and bicarbonate.

Hyperpolarized Carbon-13 Metabolic Imaging of Patients With Brain Tumors 5

极T代谢磁共振全球科研集锦

63

voxels from the T2 lesion show peaks from lactate,

pyruvate-hydrate, and pyruvate, whereas the voxels from

contralateral NAB have peaks from lactate, pyruvatehydrate, bicarbonate, and pyruvate. Note that because of

the reduced spectral bandwidth with EPSI, the bicarbonate peak wrapped into a location between pyruvatehydrate and pyruvate peaks (!179 ppm). In both cases,

lactate peaks were relatively high in the time period

between 12 and 24 seconds after the start of data acquisition. The spectra in Figure 5 are from voxels in NAB for

patients 3 to 6 at time points where the lactate was highest. Although it is difficult to compare intensities

because of variations in the reception profile of the coil,

it can be seen that the relative levels of lactate and pyruvate were similar to those for patient 2, but that the

bicarbonate peak is absent for the spectrum from patient

6. This may be attributed to the smaller voxel size and

the use of 12 rather than 10 phase encodes for data

acquisition.

Table 3 shows metabolite intensities and ratios for

spectra from patients 2 to 8 that were summed over all

time points. In all cases, the maximum lactate and bicarbonate SNRs were higher in the NAB than in the T2

lesion. Of particular interest is that the 2 patients whose

data were acquired with the multiband variable flip

angle excitation scheme (patients 7 and 8) have much

higher maximum SNR for bicarbonate in NAB (28.8 and

34.1) than patients 2 to 6, but, even in this case, the

maximum SNR of bicarbonate from the T2 lesion was

close to or lower than the level considered to be detectable (4.0 and 5.4). The maximum SNR of lactate in the

T2 lesion was variable between patients, which reflects

not only differences in its location relative to the sensitivity profile of the RF coils, but also was attributed to

differing contributions from tumor versus treatment

effects. The mean lactate/pyruvate for voxels in NAB

that had sufficient SNR to provide good estimates of

metabolite levels for patients 2 to 6 ranged from 0.18 to

0.38, whereas for patients 7 and 8 they were 0.98 and

0.84. The corresponding values for bicarbonate/pyruvate

were 0.06 to 0.15 versus 0.32 and 0.37. The increased

metabolite ratios in patients 7 and 8 are likely a result of

the multiband variable flip angle scheme used only in

these patients.

Figure 6 shows a T1 post-Gd image, maps of pyruvate,

lactate, and bicarbonate SNR, and spectral arrays that

were summed over all time points for patient 2. The low

signal in the center of the brain is attributed to both the

coil reception profile and the lower tissue content on

voxels overlapping with the ventricles.

Figure 7 illustrates representative data from patient 2

and patient 8, showing anatomical images as well as

color overlays of lactate/pyruvate and bicarbonate/pyruvate on the post-Gd T1-weighted anatomical images.

Despite the differences in acquisition parameters and

ratio values, both data sets showed substantial conversion of pyruvate to lactate in NAB and a lack of bicarbonate in the T2 lesion. Patient 2 had lactate/pyruvate in

the T2 lesion with mean value of 0.30 compared with

0.38 in their NAB, whereas patient 8 had mean lactate/

pyruvate of 0.41 within the T2 lesion compared with

0.84 in their NAB.

FIG. 4. Temporal changes in spectra from the highlighted voxels in (a) NAB and (b) the T2 lesion from patient 2. The spectra with peaks

shaded in gray are from the data reconstructed at the acquired reference frequency, and the peaks shaded in black represent the data

reconstructed at the bicarbonate frequency. The spectra shown are from time at 0, 6, 12, 18, and 24 seconds from the start of data

acquisition. The highest pyruvate and pyruvate-hydrate occur at 6 seconds, whereas the highest lactate occurs at time point 15 seconds. It can be seen that there are lactate and bicarbonate peaks in NAB and lactate peaks in the T2 lesion during the period of 9 to 24

seconds. The chemical shift range is 186.46 to 169.61ppm for all spectroscopic voxels. The intensity scales are arbitrary, but were kept

the same for all spectra.

6 Park et al.

极T代谢磁共振全球科研集锦

64

DISCUSSION

This study demonstrated the safety and feasibility of

using hyperpolarized 13C metabolic imaging for measuring real-time metabolism in the human brain utilizing

pyruvate that is transported across the blood–brain barrier by monocarboxylic transporter MCT1 (27). The slicelocalized and 2D EPSI pulse sequences and 13C coils

were first tested in a head-shaped phantom. The spatial

variation in signal intensity that was observed in the

phantom data demonstrated the anticipated variation

due to the reception profile of the 2 four-channel paddle

array coils, with a 3- to 4-fold reduction in intensity in

the center of the phantom versus voxels within 2 to 4 cm

of the lateral edges. The focus of these first patient studies was to determine the safety and feasibility. Another

focus of the initial study was to define the time course

of delivery and metabolism of hyperpolarized

[1-13C]pyruvate in both the anatomical lesion and normal

appearing brain, as opposed to seeking information about

potential diagnostic capabilities at this early stage.

The SPINlab polarizer and the associated QC system

provided pyruvate solution with an appropriate pH, temperature, EPA concentration, and polarization level. The

average polarization was of 37% which was approximately 2-fold higher than prior polarizations achieved

using pre-clinical polarizers (12). This can be explained

by the low sample vial temperature in the SPINlab

(approximately 0.8 K) compared to the preclinical system

(approximately 1.35 K) (28) and its higher magnetic field

(5T as opposed to 3.35T). Although the dissolution and

FIG. 5. FLAIR images and spectra in the highlighted NAB voxels at time with maximum lactate from (a) patient 3, (b) patient 4, (c) patient

5, and (d) patient 6. The spectra with peaks shaded in gray are from the data reconstructed at the acquired reference frequency and

the peaks shaded in black represent the data reconstructed at the bicarbonate frequency. The relative levels of pyruvate are similar, but

the lactate is slightly lower in all patients shown and the bicarbonate is not detectable for patient 6, for whom the acquisition had the

smallest voxel size 1.5 ! 1.5 ! 2 cm and 12 rather than 10 phase encodes. The chemical shift range is 186.46 to 169.61 ppm for all

spectroscopic voxels. The intensity scales are arbitrary and were manually set to have similar pyruvate intensity levels between

subjects.

Table 3

Maximum SNR of Metabolites From the Summed Spectra and Ratios of Lactate/Pyruvate and Bicarbonate/Pyruvate

Patient no.

Max SNR: NAB Max SNR: T2 lesion Mean ratios: NAB Mean ratios: T2 lesion

lac bicarb lac bicarb n lac/pyr n bicarb/pyr n lac/pyr n bicarb/pyr

2 77.7 30.4 51.6 2.0 12 0.38 12 0.15 8 0.30 8 0.02

3 45.2 18.8 8.7 5.8 20 0.31 20 0.14 0 n/a 0 n/a

4 41.5 11.8 8.5 1.5 7 0.23 7 0.07 0 n/a 0 n/a

5 39.8 10.8 4.0 2.7 13 0.22 13 0.07 0 n/a 0 n/a

6 28.0 9.2 4.4 3.9 10 0.18 10 0.06 0 n/a 0 n/a

7 76.7 28.8 24.8 4.0 16 0.98 16 0.32 2 0.58 2 0.08

8 79.1 34.1 22.8 5.4 20 0.84 20 0.37 5 0.41 5 0.07

Maximum SNR and mean ratios are from voxels that were either: normal appearing brain (NAB)—at least 80% from NAB, not overlapping with the T2 lesion and with lactate SNR >10.0 and bicarbonate SNR > 5.0; or T2 lesion—overlapped by more than 30% with the T2

lesion with lactate SNR >10.0 (n ¼ number of voxels satisfying criteria). Note that the multiband, variable flip angle scheme used for

patients 7 and 8 will alter the metabolite ratios. lac, bicarb, and pyr represent lactate, bicarbonate, and pyruvate, respectively.

n/a, not applicable.

Hyperpolarized Carbon-13 Metabolic Imaging of Patients With Brain Tumors 7

极T代谢磁共振全球科研集锦

65

ensuing QC procedure, which lasted approximately 37

seconds, extended the time from the start of dissolution

to the scan (mean time of 88 seconds) compared to the

preclinical system (typically 10–15 seconds), the high

polarization levels generated from the SPINlab system

were important for providing 13C data from human brain

FIG. 6. A post-Gd T1-weighted image, color maps of integrated pyruvate, lactate, and bicarbonate SNR and arrays of summed spectra

from patient 2. The spectra with peaks shaded in gray are from the data reconstructed at the acquired reference frequency, and the

peaks shaded in black represent the data reconstructed at the bicarbonate frequency. The lactate and pyruvate peak intensities were

determined from the former gray spectra and the bicarbonate peak intensities from the black spectra. The proposed coil setup with the

clamshell transmit and the eight-channel bilateral receive arrays allowed the acquisition of 13C signals across the majority of the brain,

but with significantly lower SNR in central regions. The low SNR in the center of the brain was accentuated by reduced signal in the

ventricles.

FIG. 7. Anatomical and interpolated hyperpolarized 13C metabolite ratio image overlays estimated from 2D EPSI dynamic data acquired at

2 cm in plane resolution and 2-cm slice thickness from 2 patients with treated glioblastoma multiforme. The differences in scales for the color

overlay images are attributed to the single-band constant flip angle excitation scheme utilized for patient 2 and the multiband variable flip

angle excitation scheme used for patient 8. Note that, in both cases, the bicarbonate/pyruvate was higher in regions of NAB. The patient in the

upper images (patient 2) had progressive tumor at the time of this exam, and the patient in the lower images (patient 8) was not characterizing

as progressing until subsequent follow-up scans. lac/pyr and bicarb/pyr represent lactate/pyruvate and bicarbonate/pyruvate, respectively.

8 Park et al.

极T代谢磁共振全球科研集锦

66

with high SNR. For patient 3, the pH was measured

manually by the pharmacist, which extended the time

from dissolution to data acquisition to 114 seconds and

resulted in the maximum SNR of pyruvate and lactate

being lower compared to other data.

The coil setup with the clamshell transmit and the

eight-channel bilateral receive arrays allowed the acquisition of 13C signals across the majority of the brain, but

with significantly lower SNR in the central regions. This

was expected based upon knowledge of the reception

profile of the paddle coils and results from the phantom

experiments, but was accentuated by the reduced signal

in voxels overlapping the ventricles. Care must be taken

when using this experimental setup to avoid the misinterpretation of the variation in the levels of metabolite

signals from different parts of the brain. When the SNR

is adequate, the use of metabolite ratios provides a more

meaningful within-subject comparison. One approach to

correct for this is to integrate all of the metabolic signals

on a voxel-by-voxel basis for each coil and to use these

as estimates of the coil sensitivity maps. Although this

provides a simple approach that does not rely upon

external standards, it may overestimate parameters in

regions with high contributions from the vasculature or

when metabolite T1s are very different. Other methods

include the use of coil reception profiles that are estimated from numerical simulations or maps obtained

from phantom studies (29), as well as the development

of 13C volume or multichannel head coils that provide

more homogenous reception profiles.

The time course of signals from the slice-localized and

2D EPSI 13C data showed a similar pattern, with the

maximum lactate appearing between 6 and 9 seconds

after the maximum pyruvate for all patients. Although

there may be differences in the arrival time of the pyruvate based upon heart rate and local vasculature, the values observed for the 2D EPSI data were within one time

point of the mean value and were similar to findings

from our previously reported studies in nonhuman primates (21). Note that the flip angle excitation scheme

employed for patients 2 to 6 was the simplest possible

(uniform across the spectrum and constant of 10 ! for

each excitation) and the number of phase encodes was

similar (10 for patients 2–5, 7, and 8 and 12 for patient

6). Although this does provide a clear picture of the

dynamic processes associated with delivery and metabolism of pyruvate within the brain, it uses more of the

available magnetization during the early time points

than the multiband, variable flip angle scheme used for

patients 7 and 8, which began with flip angles of 1.2 !

for pyruvate and 8.7 ! for lactate (22,23). The higher SNR

and ratio values that were obtained for lactate and bicarbonate using the more complex acquisition scheme indicates that it should be considered for future studies that

require improved spatial resolution and increased coverage. The incorporation of compressed sensing reconstruction (2,30,31) and frequency-specific echo planar

imaging (19) are also likely to be important for obtaining

3D 13C metabolic imaging data from the brain.

The bicarbonate signals observed in the patient studies

suggest that hyperpolarized 13C pyruvate may be useful

for probing mitochondrial metabolism. Although the SNR

of bicarbonate peaks was relatively small compared to

pyruvate and lactate, it was detected in NAB for 7 of 8 subjects. Of particular interest is that it was clearly present in

voxels from contralateral brain, but absent in voxels from

the T2 lesion for patients 2, 7, and 8, whose scans showed

sufficiently high enough SNR and had large enough

lesions to provide definitive results. This differential is of

interest for future studies and needs to be considered in

pulse-sequence design. The relatively low metabolite signals in voxels overlapping with the T2 lesion from

patients 3 to 6 are consistent with them either being in

regions with low reception profile or corresponding

mainly to treatment effects rather than residual or recurrent tumor. None of these voxels were enhancing on corresponding post-Gd T1-weighted images or were in portions

of the lesion that were found to progress in subsequent

scans. Although the relatively high conversion of pyruvate

to lactate in NAB meant that there was not a clear differential for voxels in the T2 lesion and surrounding brain for

patient 2, who was the only subject with clinical status

defined as progressive tumor and had high enough lactate

SNR in tumor, it does suggest that this technique may be

of interest for studying changes in metabolism associated

with other neurological diseases.

The purpose of this paper was to report upon initial

patient studies and to introduce the experimental setup

that was designed for the acquisition of 13C data from

the human brain. Although a relatively small group of

patients were included, the time course of changes in

metabolite levels and the appearance of lactate and bicarbonate in NAB were consistent between subjects and

provide a basis for designing future, more advanced data

acquisition schemes and experimental setups. The observation of substantial conversion of pyruvate to lactate in

normal brain is in contrast to our previous results in

studies from normal rats and nonhuman primates

(2,12,21), where the conversion of pyruvate to lactate

was relatively low. In our previous rodent studies that

used similar experimental conditions, the bicarbonate

signal was not detected (12,13); however, there have

been several other reports in rats that have demonstrated

the detection of bicarbonate in normal brain as well as in

glioma (32–34). Whether this was attributed to differences in brain metabolism between humans and other

species or because of the anesthesia (35) is unclear. Of

interest for evaluating patients with brain tumors is that

no bicarbonate was detected in lesions corresponding to

recurrent tumor, but that the ratios of lactate/pyruvate in

T2 lesion were similar to or lower than the lactate/pyruvate in NAB. Further technical studies are required to

optimize data acquisition parameters for the brain to provide 3D coverage and finer spatial resolution. This will

be important for using hyperpolarized 13C agents to characterize metabolism in normal gray and white matter, as

well as detecting changes associated with brain tumors

and other types of pathology.

CONCLUSIONS

Experimental strategies for implementing hyperpolarized

13C metabolic imaging in the human brain have been

developed, and initial patient studies have confirmed

Hyperpolarized Carbon-13 Metabolic Imaging of Patients With Brain Tumors 9

极T代谢磁共振全球科研集锦

67

the safety and feasibility of using this technology. The

results obtained indicate that hyperpolarized

[1-13C]pyruvate was transported across the blood–brain

barrier and was converted to metabolic products,

[1-13C]lactate and 13C-bicarbonate, in a time frame that

can be measured using the pulse sequences and RF coils

designed for this study. Although additional optimization is required, these initial findings support further

investigation of the technology in patients with brain

tumors and other neurological diseases.

ACKNOWLEDGMENTS

The authors gratefully acknowledge the assistance of Jennifer Chow, Romelyn Delos Santos, Adam Autry, Kimberly Okamoto, RN, and Mary Mcpolin, RT, for assisting

in patient scans. The first author was supported by an

NCI training grant in translational brain tumor research

(T32 CA151022), Kure It Grant for Underfunded Cancer

Research, Discovery Grant from American Brain Tumor

Association, and the National Research Foundation

(NRF) of Korea grant funded by Ministry of Science and

ICT (No. 2017R1C1B5018396).

REFERENCES

1. Ardenkjaer-Larsen JH, Fridlund B, Gram A, et al. Increase in signalto-noise ratio of > 10,000 times in liquid-state NMR. Proc Natl Acad

Sci U S A 2003;100:10158–10163.

2. Park I, Hu S, Bok R, et al. Evaluation of heterogeneous metabolic profile in an orthotopic human glioblastoma xenograft model using compressed sensing hyperpolarized 3D 13C magnetic resonance

spectroscopic imaging. Magn Reson Med 2013;70:33–39.

3. Brindle K. New approaches for imaging tumour responses to treatment. Nat Rev Cancer 2008;8:94–107.

4. Warburg O. On the origin of cancer cells. Science 1956;123:309–314.

5. Albers MJ, Bok R, Chen AP, et al. Hyperpolarized 13C lactate, pyruvate, and alanine: noninvasive biomarkers for prostate cancer detection and grading. Cancer Res 2008;68:8607–8615.

6. DeVience SJ, Lu X, Proctor J, et al. Metabolic imaging of energy

metabolism in traumatic brain injury using hyperpolarized [1-

13C]pyruvate. Sci Rep 2017;7:1907.

7. Golman K, Petersson JS, Magnusson P, et al. Cardiac metabolism

measured noninvasively by hyperpolarized 13C MRI. Magn Reson

Med 2008;59:1005–1013.

8. Guglielmetti C, Najac C, Didonna A, Van der Linden A, Ronen SM,

Chaumeil MM. Hyperpolarized (13)C MR metabolic imaging can

detect neuroinflammation in vivo in a multiple sclerosis murine

model. Proc Natl Acad Sci U S A 2017;114:E6982–E6991.

9. Hu S, Balakrishnan A, Bok RA, et al. 13C-pyruvate imaging reveals

alterations in glycolysis that precede c-Myc-induced tumor formation

and regression. Cell Metab 2011;14:131–142.

10. Kurhanewicz J, Vigneron DB, Brindle K, et al. Analysis of cancer

metabolism by imaging hyperpolarized nuclei: prospects for translation to clinical research. Neoplasia 2011;13:81–97.

11. Mignion L, Dutta P, Martinez GV, Foroutan P, Gillies RJ, Jordan BF.

Monitoring chemotherapeutic response by hyperpolarized 13Cfumarate MRS and diffusion MRI. Cancer Res 2014;74:686–694.

12. Park I, Larson PE, Zierhut ML, et al. Hyperpolarized 13C magnetic

resonance metabolic imaging: application to brain tumors. Neurooncology 2010;12:133–144.

13. Park I, Mukherjee J, Ito M, et al. Changes in pyruvate metabolism

detected by magnetic resonance imaging are linked to DNA damage

and serve as a sensor of temozolomide response in glioblastoma cells.

Cancer Res 2014;74:7115–7124.

14. Nelson SJ, Kurhanewicz J, Vigneron DB, et al. Metabolic imaging of

patients with prostate cancer using hyperpolarized [1-(1)(3)C]pyruvate. Sci Transl Med 2013;5:198ra108.

15. Cunningham CH, Lau JY, Chen AP, et al. Hyperpolarized 13C Metabolic MRI of the Human Heart: Initial Experience. Circ Res 2016;119:

1177–1182.

16. Tropp J, Lupo JM, Chen A, et al. Multi-channel metabolic imaging,

with SENSE reconstruction, of hyperpolarized [1-(13)C] pyruvate in a

live rat at 3.0 tesla on a clinical MR scanner. J Magn Reson 2011;208:

171–177.

17. Ohliger MA, Larson PE, Bok RA, et al. Combined parallel and partial

fourier MR reconstruction for accelerated 8-channel hyperpolarized

carbon-13 in vivo magnetic resonance Spectroscopic imaging (MRSI).

J Magn Reson Imaging 2013;38:701–713.

18. Stollberger R, Wach P. Imaging of the active B1 field in vivo. Magn

Reson Med 1996;35:246–251.

19. Gordon JW, Vigneron DB, Larson PE. Development of a symmetric

echo planar imaging framework for clinical translation of rapid

dynamic hyperpolarized 13 C imaging. Magn Reson Med 2017;77:

826–832.

20. Larson PE, Bok R, Kerr AB, et al. Investigation of tumor hyperpolarized [1-13C]-pyruvate dynamics using time-resolved multiband RF

excitation echo-planar MRSI. Magn Reson Med 2010;63:582–591.

21. Park I, Larson PE, Tropp JL, et al. Dynamic hyperpolarized carbon-13

MR metabolic imaging of nonhuman primate brain. Magn Reson Med

2014;71:19–25.

22. Larson PE, Hu S, Lustig M, et al. Fast dynamic 3D MR spectroscopic

imaging with compressed sensing and multiband excitation pulses

for hyperpolarized 13C studies. Magn Reson Med 2011;65:610–619.

23. Xing Y, Reed GD, Pauly JM, Kerr AB, Larson PE. Optimal variable

flip angle schemes for dynamic acquisition of exchanging hyperpolarized substrates. J Magn Reson 2013;234:75–81.

24. Crane JC, Olson MP, Nelson SJ. SIVIC: Open-Source, Standards-Based

Software for DICOM MR Spectroscopy Workflows. Int J Biomed Imaging 2013;2013:169526.

25. Nelson SJ. Analysis of volume MRI and MR spectroscopic imaging

data for the evaluation of patients with brain tumors. Magn Reson

Med 2001;46:228–239.

26. Stollberger R, Wach P. Imaging of the active B1 field in vivo. Magn

Reson Med 1996;35:246–251.

27. Halestrap AP. Monocarboxylic acid transport. Compr Physiol 2013;3:

1611–1643.

28. Ardenkjaer-Larsen JH, Leach AM, Clarke N, Urbahn J, Anderson D,

Skloss TW. Dynamic nuclear polarization polarizer for sterile use

intent. NMR Biomed 2011;24:927–932.

29. Dominguez-Viqueira W, Geraghty BJ, Lau JY, Robb FJ, Chen AP,

Cunningham CH. Intensity correction for multichannel hyperpolarized 13C imaging of the heart. Magn Reson Med 2016;75:859–865.

30. Hu S, Lustig M, Balakrishnan A, et al. 3D compressed sensing for

highly accelerated hyperpolarized (13)C MRSI with in vivo applications to transgenic mouse models of cancer. Magn Reson Med 2010;

63:312–321.

31. Hu S, Lustig M, Chen AP, et al. Compressed sensing for resolution

enhancement of hyperpolarized 13C flyback 3D-MRSI. J Magn Reson

2008;192:258–264.

32. Butt SA, Sogaard LV, Magnusson PO, et al. Imaging cerebral 2-

ketoisocaproate metabolism with hyperpolarized (13)C magnetic resonance spectroscopic imaging. J Cereb Blood Flow Metab 2012;32:

1508–1514.

33. Park JM, Recht LD, Josan S, et al. Metabolic response of glioma to

dichloroacetate measured in vivo by hyperpolarized (13)C magnetic

resonance spectroscopic imaging. Neuro-oncology 2013;15:433–441.

34. Park JM, Spielman DM, Josan S, et al. Hyperpolarized (13)C-lactate to

(13)C-bicarbonate ratio as a biomarker for monitoring the acute

response of anti-vascular endothelial growth factor (anti-VEGF) treatment. NMR Biomed 2016;29:650–659.

35. Josan S, Hurd R, Billingsley K, et al. Effects of isoflurane anesthesia

on hyperpolarized (13)C metabolic measurements in rat brain. Magn

Reson Med 2013;70:1117–1124.

SUPPORTING INFORMATION

Additional supporting information can be found in the online version of this

article.

Fig. S1. Variable flip angle scheme for pyruvate and lactate. The flip angles

for bicarbonate were the same as for lactate.

10 Park et al.

极T代谢磁共振全球科研集锦

68

Metabolic Imaging of Patients with Prostate Cancer Using Hyperpolarized [1-13C]

Pyruvate

研究背景

研究对象

研究过程

Ԩ๯ْට༹ׯၟ研৯ೠࠚକג极ࣅ] 1-13C] եཛྷ໗ፕྺᅃዖݥ෇ൔႠ՗ኙമଚ၇ҵ࣒ኁ዗ୀ代谢߀Վڦᄱ࿿ڦҾ全Ⴀ

ࢅ੗ႜႠă၎ܔᇀدཥڦ磁共振 (MR) ׯၟLj๑ᆩג极ࣅवܔิంဣཥ৊ႜׯ੗ၟᅜ๑႑ࡽሺഽ 10,000 Ԡᅜฉă঳

ࢇ ా༹13C MR ຕ਍ڦ֑໏੺集Lj੗ᅜሞब௱ዓాೠࠚ] 1-13C] եཛྷ໗तഄ代谢ׂ࿿ළ໗Ăեҽ໗ࢅ༐໗ൠჸڪ࿿ዊڦ

ݴքăҵኢఇ႙ڦଣضമ研৯ॠ֪ڟ዗ୀዐג极ڦࣅ] 1-13C] ළ໗຤ೝืߛLj[1-13C] ළ໗ /[1-13C] եཛྷ໗ڦԲ૩ሞߛप

՚዗ୀዐืߛLjࠀׯዎଐࢫইگăཚ߀ࡗሰׂิג极ࣅवڦᅏഗLjࠓॺጆோڦพೕ၍ංઠॠ֪ 13C ᇱጱࢃLjᅜतਸ

݀ႎڦஞ؋Ⴞଚᅜᆶၳվइ႑ࡽLjॽኄၜरຍገࣅၠටૌă

31 ఁ࿄ঢ়ዎଐĂऄॠኤํڦਆ၌Ⴀമଚ၇ҵ࣒ኁ

മଚ၇ҵ๟ፌ׉९ڦҵኢኮᅃLjெࡔ౎௅Ԓڦߢႎթ૩ࡗג 200,000ăໜጣ๑ᆩეൣമଚ၇༬ᅴႠੇᇱ (PSA) ࢅકቛ

ఇӱঢ়኱ג׌ำ (TRUS) ᆅڦڞऄॠ৊ႜෲֱڦሺेLjമଚ၇ҵ࣒ኁሞራ೺൐੗ీ߸ᅟዎଐڦ঩܎ዐԥඓۨă

ݛڦ๴থݘ࠽ுᆶԥ࣏ణമ൐ኵᆶ၌Ljժॏڦ௬ݛࢫᇨࠚཥरຍሞೠدڍമଚ၇ҵLjࠚԥᆩᇀೠၟׯظ࿮๑ᆩ࠶৑

๕ઠ༵ࠃᆶ࠲ൔလႠࢅዎଐݒᆌڦ႑တă

ג极ࣅ13C MRI ๟ᅃዖႎ႙ݴጱׯၟरຍLjഄ႑ࡽഽ܈മ໯࿄ᆶں༵ߛକ 10,000 ዁ 100,000 ԠLj੗ᆩᇀ֪॔ాᇸႠิ

代谢ăࢅฝൽڦጱݴ࿿

13C ຕ਍ڦ֑集੗ᅜ๑ᆩฆᆩ MR ෢௮ᅏLj঳ࢇጆோڦஞ؋Ⴞଚࢅሞ๢ړೕ୲݀ໃࢅথ๭ڦพೕ၍ංઠํ၄ă

Ԩ๯ْට༹研৯ڦዷᄲణՔ๟ኤ௽ג极ࣅ] 1-13C] եཛྷ໗ጀพܔമଚ၇ҵఴႠڦҾ全Ⴀࢅ੗ႜႠăሞඓۨ੗ᅜҾ全ڿ

ໃڦෙ߲वଉ຤ೝڦፌۇٴኵࢫLjْᄲణڦ๟ೠڿࠚໃ዁മଚ၇ۯڦ૰ბժೠࠚҵኢ൶ᇘᇑഄ໱൶ᇘڦ] 1-13C] ළ໗/

[1-13C] եཛྷ໗ڦֶᅴ࿚༶ăࠀׯኤ௽޿रຍ੗ᆌᆩᇀටૌLjࣷྺॽഄ૧ᆩᇀॠ֪ݴࢅ೺ҵኢĂॠ֪዗ୀ৊ቛࢅ֪॔

ዎଐݒᆌ༵ࠃऐࣷă

极T代谢磁共振全球科研集锦

69

研究结果

结 论

应用方向

ג极ࣅ] 1-13C] ළ໗ /[1-13C] եཛྷ໗ڦዡၠ T2 े඄཮ၟ

ࢅे۠ઠጲՔऻྺ B ڟ D ڦෙఁ࣒ኁă໯ᆶෙఁ࣒ኁ

਩࣒ᆶঢ়ऄॠኤํڦ ෧૛߭3+3 पമଚ၇ҵ ժথ๴ፌ

ߛवଉגڦ极ࣅ] 1-13C] եཛྷ໗ (0.43 ml/kg)ă ࣒ኁ B ڦ

ړമ PSA ྺ 5.1ng/mlLj࣒ኁ C ڦ PSA ྺ 9.8 ng/mlLj࣒

ኁ D ڦ PSA ྺ 1.9ng/mlă

ҵဦԇڦႎר代谢ᇑॳ੃ဦԇփཞăਏ༹ઠຫLj໲்Բኟڦ׉ཞૌၩࡼ߸ڦܠեཛྷ໗Ljኄ๟༛ট঴࠲ڦ॰ݴׯăᅺُLj

Nelson तഄཞ๚๑ᆩג极ࣅႚ๕ڦᙽᘘ໗DŽ[1-13C] եཛྷ໗Džܔ] ࿿ׂഄ1-13C] ළ໗ڦሺे຤ೝᅜतեཛྷ໗ገࣅྺළ໗

ڦཚଉ৊ႜ௺ׯߌၟă [1-13C] եཛྷ໗ჸ๬वሞኄ૛ᆩᇀമଚ၇ҵఴႠڦْ๯ට༹研৯ă

࣒ኁথ๴କփཞҾ全वଉڦ] 1-13C] եཛྷ໗ăᆩג极ࣅ 13C MR ܔኄၵ࣒ኁ৊ႜ੺໏ׯၟLjీࠕሞڇ้߲क़ۅ༵ۯࠃༀ

ࢅ 3D ຕ਍ăሞ໯ᆶऄॠኤํྺҵኢ࣒ڦኁዐ਩ॠ֪ڟ዗ୀăܸ൐Ljዘᄲڦ๟Ljཚࡗ 13C MRILjፕኁీࠕሞ๑ᆩഄ໱

ҵኢăڟੂᇘዐ൶മଚ၇ڦୀ዗؜ֱ࿄ॠ݆ݛၟׯ೪঴ࡀ׉

ᆯᇀీࠕҾ全ܔں዗ୀ࿋ዃ৊ႜׯၟLjժໜ้क़ڦླྀᅎ߶ጷ዗ୀ代谢Ljג极ࣅ 13C MRI ੗ᆩᇀ؛๔ኑ܏ࢅ֪॔ዎଐă

৑࠶Ԩ研৯ዐ࣒ڦኁ࣒ᆶራ೺दթLjڍፕኁණྺ [1-13C] ළ໗ /[1-13C] եཛྷ໗ཚଉኻࣷໜጣ዗ୀݴपܸሺेLj๑ኄዖ

ׯၟरຍ๢ᆩᇀ߸ྟ೺ࢅ߸ਏൔလႠڦҵኢă࿄ઠڦ研৯ॽ֨ዘᇀᆫࣅᄱ࿿዆ԢڿࢅໃLjᅜඓԍኄዖׯၟरຍ੗ᅜ

๑໯ᆶଣ࣍ضৣዐ࣒ڦኁ๴ᅮă

മଚ၇ҵ

极T代谢磁共振全球科研集锦

70

IMAGING

Metabolic Imaging of Patients with Prostate Cancer

Using Hyperpolarized [1-13C]Pyruvate

Sarah J. Nelson,1,2* John Kurhanewicz,1 Daniel B. Vigneron,1,2 Peder E. Z. Larson,1

Andrea L. Harzstark,3 Marcus Ferrone,4 Mark van Criekinge,1 Jose W. Chang,4 Robert Bok,1

Ilwoo Park,1 Galen Reed,1 Lucas Carvajal,1 Eric J. Small,3 Pamela Munster,3 Vivian K. Weinberg,5

Jan Henrik Ardenkjaer-Larsen,6 Albert P. Chen,6 Ralph E. Hurd,6 Liv-Ingrid Odegardstuen,6

Fraser J. Robb,7 James Tropp,6 Jonathan A. Murray6

This first-in-man imaging study evaluated the safety and feasibility of hyperpolarized [1-13C]pyruvate as an agent

for noninvasively characterizing alterations in tumor metabolism for patients with prostate cancer. Imaging living

systems with hyperpolarized agents can result in more than 10,000-fold enhancement in signal relative to

conventional magnetic resonance (MR) imaging. When combined with the rapid acquisition of in vivo 13C MR data,

it is possible to evaluate the distribution of agents such as [1-13C]pyruvate and its metabolic products lactate,

alanine, and bicarbonate in a matter of seconds. Preclinical studies in cancer models have detected elevated

levels of hyperpolarized [1-13C]lactate in tumor, with the ratio of [1-13C]lactate/[1-13C]pyruvate being increased

in high-grade tumors and decreased after successful treatment. Translation of this technology into humans was

achieved by modifying the instrument that generates the hyperpolarized agent, constructing specialized radio

frequency coils to detect 13C nuclei, and developing new pulse sequences to efficiently capture the signal. The

study population comprised patients with biopsy-proven prostate cancer, with 31 subjects being injected with

hyperpolarized [1-13C]pyruvate. The median time to deliver the agent was 66 s, and uptake was observed about

20 s after injection. No dose-limiting toxicities were observed, and the highest dose (0.43 ml/kg of 230 mM agent)

gave the best signal-to-noise ratio for hyperpolarized [1-13C]pyruvate. The results were extremely promising in

not only confirming the safety of the agent but also showing elevated [1-13C]lactate/[1-13C]pyruvate in regions of

biopsy-proven cancer. These findings will be valuable for noninvasive cancer diagnosis and treatment monitoring

in future clinical trials.

INTRODUCTION

Prostate cancer is one of the most common cancers, with more than

200,000 new cases being reported annually in the United States (1).

Owing to increased screening using serum prostate-specific antigen

(PSA) and extended-template transrectal ultrasound (TRUS)–guided

biopsies, patients with prostate cancer are being identified at an earlier

and potentially more treatable stage (2). Once detected, the decision

on how to manage prostate cancer poses a dilemma because there is a

tremendous range in biologic diversity. They are treated with a broad

spectrum of approaches from “active surveillance” to more aggressive

surgical, radiation-based, and other focal therapies (3, 4). Such therapies

have trade-offs because, no matter how well they are delivered, there

can be changes in health-related quality of life (5). In practice, many

prostate cancers follow an indolent course that would not threaten the

duration or quality of lives for the affected men, but the natural history

of individual tumors is difficult to predict using currently available

prognostic data (6, 7). Conversely, between 22 and 35% of men presenting with clinically advanced prostate cancer, who are treated with what

was thought to be definitive radiation or surgery, suffer a posttreatment

biochemical recurrence (8). The ability to predict outcome for individual patients and thereby select the most appropriate treatment is

a critically important, but so far unmet, clinical need.

Although noninvasive imaging is used to assess prostate cancer,

conventional techniques have limited value for assessing prognosis,

and there is no widely accepted modality that provides information

about aggressiveness and response to therapy (9). Proton magnetic

resonance spectroscopic imaging (1

H MRSI) has been applied to assess

the metabolic properties of localized prostate cancer and, although it

has shown clear advantages over anatomic imaging alone (10), is

limited by its relatively low spatial resolution and acquisition time,

which is in the range of 10 to 20 min (11). 18F-Fluorodeoxyglucose

positron emission tomography provides information about increased

glucose uptake and phosphorylation, but has been shown to inadequately evaluate the presence and aggressiveness of prostate

cancer because of both its relatively low uptake and collection in

the bladder (12).

Hyperpolarized 13C MRI is a new molecular imaging technique with

an unprecedented gain in signal intensity of 10,000- to 100,000-fold (13)

that can be used to monitor uptake and metabolism of endogenous

biomolecules (14, 15). The magnitude of the increase in sensitivity

depends on the degree of polarization that is achieved, the T1 relaxation time of the 13C agent, the delivery time, and the MR methods

applied. Hyperpolarized agents are generated by mixing 13C-labeled

compounds with an electron paramagnetic agent (EPA), placing them

in a 3.35-T magnetic field, cooling to ~1 K, and using microwaves to

1

Surbeck Laboratory of Advanced Imaging, Department of Radiology and Biomedical

Imaging, University of California, San Francisco, San Francisco, CA 94158, USA. 2Department of Bioengineering and Therapeutic Sciences, University of California, San Francisco,

San Francisco, CA 94158, USA. 3

Department of Medicine, University of California, San

Francisco, San Francisco, CA 94143, USA. 4Department of Clinical Pharmacy, University of

California, San Francisco, San Francisco, CA 94143, USA. 5

Department of Biostatistics,

University of California, San Francisco, San Francisco, CA 94143, USA. 6

General Electric

Healthcare, Waukesha, WI 53188, USA. 7

USA Instruments Inc., Aurora, OH 44202, USA.

*Corresponding author. E-mail: sarah.nelson@ucsf.edu

RESEARCH ARTICLE

www.ScienceTranslationalMedicine.org 14 August 2013 Vol 5 Issue 198 198ra108 1

by guest on March 5, 2018 http://stm.sciencemag.org/ Downloaded from

极T代谢磁共振全球科研集锦

71

transfer polarization from the electron spin of the EPA to the 13C

nuclei of the biomolecule (13). Once the polarization has reached

the required level, the sample is rapidly dissolved with hot, sterile water and neutralized to physiological pH, temperature, and osmolarity.

Intravenous injection of the hyperpolarized solution and observation

using 13C MR allow its delivery and metabolic products to be monitored (15). The data must be obtained as rapidly as possible after dissolution because the enhancement decays at a rate determined by the

T1 relaxation time of the agent, which is about 60 s for [1-13C]pyruvate

at 3 T. Translation of hyperpolarized technology into human subjects

has been challenging because it requires specialized instrumentation to

prepare the agent in a sterile environment, filter out the EPA, perform

quality control (QC), and rapidly deliver samples to the patient.

The acquisition of 13C data can be achieved using commercially

available MR scanners in conjunction with specialized pulse sequences

(16) and radio frequency coils designed to transmit and receive at

the appropriate frequency. Increased signal from hyperpolarized

[1-13C]lactate has been observed in preclinical models of prostate

cancer relative to normal tissues, owing to both increased uptake of

[1-13C]pyruvate via monocarboxylate transporters 1 and 4 (MCT1

and MCT4) and increased expression of LDH-A and activity of lactate

dehydrogenase (LDH) (17). Levels of hyperpolarized [1-13C]lactate

and the flux of [1-13C]pyruvate to [1-13C]lactate increase with cancer

progression (pathologic grade) (18) and reduce after therapy (19, 20).

The primary goal of this first-in-human study was to demonstrate

the safety and feasibility of hyperpolarized [1-13C]pyruvate injections

in men with prostate cancer. After establishing the maximum of three

dose levels that could be safely delivered, the secondary aims were to

evaluate the kinetics of delivery to the prostate and assess differences

in [1-13C]lactate/[1-13C]pyruvate for regions of cancer versus other

tissues. Successfully demonstrating that the technology can be applied

to humans provides the opportunity to use it for detecting and staging

cancer, as well as detecting tumor progression and monitoring response to therapy.

RESULTS

Study design: Doses delivered and data acquisition

There were 31 patients with untreated, biopsy-proven localized prostate

cancer who received an injection of hyperpolarized [1-13C]pyruvate.

Their median age was 63 years (range, 45 to 75), median PSA was

5.9 ng/ml (range, 1.88 to 20.2), and median LDH was 141 IU/liter

(range, 109 to 261). Twenty-three of them had a diagnosis of Gleason

grade 6 tumor, 6 had Gleason grade 7 tumor, and 2 had Gleason grade

8 tumor. The initial (phase 1) component of the study used a standard

dose escalation design (Table 1), with six subjects being evaluated at

each of three ascending doses. Dose levels were chosen on the basis

of the range that had been shown to be safe in previous studies that

did not include imaging but did inject nonlabeled pyruvate at the

same concentrations, pH, and delivery rate (Supplementary Methods).

At each dose, three patients were scanned with a dynamic 13C sequence

that provided localization to one-dimensional (1D) slices through the

prostate, and three patients were scanned with a 13C sequence that

provided 2D or 3D spatial localization at a single time point. An additional 13 subjects were evaluated at the highest dose level in phase 2

with a mixture of dynamic and single–time point 13C MR sequences.

The specialized 13C volume transmit coil and dual 1

H/13C endorectal

coil that were designed and built for this study are shown in Fig. 1, A

to C. These functioned well and allowed 1

H images and 13C spectral

data to be acquired from all subjects.

Hyperpolarized pyruvate polarization and delivery

For the 31 samples injected into patients, the average polarization was

17.8% (range, 15.9 to 21.1), pH was 7.6 (range, 7.3 to 8.0), temperature

was 32.4°C (range, 28.8 to 36.4), and volume was 51.9 ml (range, 31.9

to 53.5). QC criteria were defined as follows: polarization to be not less

than 15%, pH in the range of 6.7 to 8.0, sample temperature in the

range of 25° to 37°C, and residual EPA concentration no higher than

3.0 mg. The mean times for getting the sample to the patient are seen

schematically in Fig. 1D: the dissolution took an average of 17.8 s

(range, 5 to 30), the QC process 13.1 s (range, 10 to 19), delivery

through the hatch into the scan room 21.8 s (range, 11 to 30), and

injection 14.9 s (range, 6 to 28). Overall, this gave an average of 67.6 s

(range, 43 to 88) to deliver the agent to the subject. The parameters for

individual patients are given in table S1. The mean injection volumes

for the patients studied at each dose were 11.8 ml (range, 10 to 14),

26.8 ml (range, 22 to 33), and 34.5 ml (range, 29 to 46), with mean

injection times of 8.5 s (range, 6 to 10), 14.8 s (range, 10 to 27), and

12.5 s (range, 1 to 28), respectively. Variation in injection times reflected

differences in the volume delivered, as well as in the time for drawing the

material from the drug product vessel into a syringe and performing the

manual injection.

Patient toxicities

Vital signs were monitored before and immediately after the imaging

examination, with subsequent telephone follow-ups over a period of

7 days to check for evidence of adverse events. In phase 1 of the study,

there were a total of 10 adverse events in eight patients (table S2).

These were all considered mild events and were classified as grade

1 by Common Terminology Criteria for Adverse Events (CTCAE)

v4.0.criteria (21). The highest dose of [1-13C]pyruvate (0.43 ml/kg)

was selected for further study on the basis of the higher signal-to-noise

ratio (SNR) of hyperpolarized [1-13C]pyruvate that was observed. In

phase 2, there were an additional 10 events observed in five patients,

but, again, none of them were considered to be dose-limiting toxicities

Table 1. Summary of study design. Number of patients, doses of

hyperpolarized [1-13C]pyruvate given, and types of MR examinations

performed in each component of the study are provided.

Study component Dose

(ml/kg) n patients Type of MRacquisition

Phase 1: Dose escalation 0.14 3 1D dynamic MRSI

3 3D MRSI

0.28 3 1D dynamic MRSI

3 3D MRSI

0.43 3 1D dynamic MRSI

3 2D MRSI

Phase 2: Refining MR methods 0.43 5 2D dynamic MRSI

3 2D MRSI

5 3D MRSI

RESEARCH ARTICLE

www.ScienceTranslationalMedicine.org 14 August 2013 Vol 5 Issue 198 198ra108 2

by guest on March 5, 2018 http://stm.sciencemag.org/ Downloaded from

极T代谢磁共振全球科研集锦

72

(DLTs). The single episode of dizziness that was seen in one patient

during phase 2 was attributed to extra dosing of atenolol, which

was used by the subject to reduce anxiety rather than the hyperpolarized agent. There was one episode of grade 2 diarrhea reported

in the phase 2 component, which was attributed to an enema that

the patient received.

1D dynamic MRSI

The purpose of the 1D spatially localized

dynamic data was to establish the time

course of delivery of the agent. The acquisition provided spectra from an axial

slab that covered the prostate and surrounding tissues and applied echo planar

encoding from slices in the right-left direction at a 3-s time resolution. These dynamic data demonstrated reproducible

delivery of hyperpolarized [1-13C]pyruvate

to the prostate and its conversion to hyperpolarized [1-13C]lactate. Figure 2A

shows a representative scan from a patient with PSA of 12.2 ng/ml, who had

a small volume of biopsy-proven prostate

cancer (Gleason grade 4 + 3) in the left

midgland and received the lowest dose of

0.14 ml/kg. Representative 13C spectra

from the same patient taken from a

slice including the tumor 36 s after injection demonstrated peaks corresponding

to [1-13C]pyruvate [173 parts per million

(ppm)] and [1-13C]lactate (185 ppm)

(Fig. 2B). Spectra from the contralateral

side of the prostate demonstrated only

[1-13C]pyruvate. A plot of spectral peak

heights from the slice overlapping the tumor is shown in Fig. 2C and from the

slice on the contralateral side of the gland

Injection

Agent in

polarizer

at 1.2 K

Polarization

Agent in

patient Dissolution QC system Hatch

D

A B

17.8 s 13.1 s 21.8 s 14.9 s

13C transmit coil

13C receive coil

1H receive coil

C

Fig. 1. 13C coil setup and the schematic steps for the delivery of hyperpolarized [1-13C]pyruvate.

(A to C) 13C transmit coil (A) and endorectal 1

H/13C receiver coil (B) used for acquiring data. The location

of the coils is outlined on (A) and (B), with the layers inside the endorectal coil being seen in (C). The

dimensions of the elements in the endorectal coil were 4 inches × 1 inches, with the total length of the

coil being 12 inches. (D) Steps involved in transferring the hyperpolarized agent from the polarizer to

the patient, and mean times required for each of them.

0 50 100 150 200 0

25

50

75

100

125

Time (s)

Signal to noise

Pyruvate

13C1

Lactate

13C1

NADH NAD+

LDH

185 180 170 ppm 175

Cancer Pyruvate

Lactate

Normal

A B C

0 50 100 150 200 0

25

50

75

100

125

Signal to noise

D

Pyruvate

Lactate

Prostate cancer region Contralateral prostate region

Time (s)

Fig. 2. 1D 13C dynamic MRSI data. Images are from a representative

patient with a current PSA of 12.2 ng/ml, a small volume of biopsy-proven

Gleason grade 4 + 3 prostate cancer in the left midgland, and who received the lowest dose (0.14 ml/kg) of hyperpolarized [1-13C]pyruvate. (A)

Axial T2-weighted image showing slices (dashed lines) obtained from 1D

spectral localization. The slice that overlaps the left prostatic peripheral

zone (right side of image) contained a small focus of reduced T2 signal

intensity corresponding to the region of biopsy-proven cancer (red arrows).

The slice overlapping the right peripheral zone (left side of image) contains

only normal prostate tissue. (B) Flux of [1-13C]pyruvate to [1-13C]lactate

catalyzed by LDH (top). Dynamic 13C spectra were obtained from the

same patient in (A) at 36 s after injection of hyperpolarized [1-13C]pyruvate

(bottom). The cancer spectrum demonstrated a lactate SNR of 25 owing to

a high flux of hyperpolarized [1-13C]pyruvate to [1-13C]lactate. (C) Plot of 1D

localized dynamic hyperpolarized pyruvate and lactate data from the slice

that overlapped the region of prostate cancer. (D) Plot of 1D localized

dynamic hyperpolarized pyruvate and lactate data from the slice that

overlapped a contralateral region of the prostate.

RESEARCH ARTICLE

www.ScienceTranslationalMedicine.org 14 August 2013 Vol 5 Issue 198 198ra108 3

by guest on March 5, 2018 http://stm.sciencemag.org/ Downloaded from

极T代谢磁共振全球科研集锦

73

in Fig. 2D. In both cases, [1-13C]pyruvate arrived at about 20 s and

reached a maximum at 24 s. [1-13C]Lactate reached a maximum signal

plateau at about 30 s after the end of injection.

For the patients who had unambiguous regions of tumor, the maximum SNRs of [1-13C]pyruvate and [1-13C]lactate in slices containing

tumor were 101.3 (range, 31.6 to 192.1) and 19.5 (range, 12.5 to 33.5),

respectively. Fitting the time course of changes in peak heights using a

previously published two-compartment kinetic model (22) resulted in

an estimated mean T1 of [1-13C]pyruvate and [1-13C]lactate of 29.2 ±

5 s and 25.2 ± 5 s (SD), respectively, and a mean [1-13C]pyruvate to

[1-13C]lactate conversion rate constant of 0.013 ± 0.003 s−1 (SD)

(range, 0.009 to 0.016).

2D dynamic MRSI

A concern with the 1D localized dynamic MRSI data was that the

contribution from hyperpolarized signals in tissues outside the prostate could confound the interpretation of estimated parameters.

Therefore, five patients in phase 2 were studied with 2D spatially localized dynamic data. One of these patients had a PSA of 3.6 ng/ml

and biopsy-proven prostate cancer in the left apex (Gleason grade 3 + 4)

with an associated focus of reduced signal intensity on the anatomic

images obtained during the MR staging examination (Fig. 3A). Hyperpolarized [1-13C]pyruvate and [1-13C]lactate from voxels in the prostate (Fig. 3B), the tumor (Fig. 3C), and a vessel outside the prostate

(Fig. 3D) demonstrated a similar time course of delivery of the agent.

There was increased conversion of [1-13C]pyruvate to [1-13C]lactate in

the tumor (Fig. 3C), with the [1-13C]pyruvate reaching a maximum by

18 ± 4 s and the [1-13C]lactate reached a maximum at 27 ± 2 s after

the start of acquisition, which was 5 s after the end of injection. After

adjusting for additional delay, the timings of maximal signal are

similar to those for the 1D dynamic data. Overall, the rate constant

(mean ± SD) was 0.045 ± 0.025 s−1 for [1-13C]pyruvate to [1-13C]lactate

in tumor voxels and 0.009 ± 0.003 s−1 for voxels coming from regions

that included blood vessels.

Single–time point MRSI

The purpose of the single–time point

spatially localized data was to compare

the relative levels of [1-13C]lactate and

[1-13C]pyruvate in regions of tumor versus normal prostate and surrounding tissue.

For the phase 1 study, the MRSI acquisitions were initially started at about 30 s

after the end of the injection. The maximum SNR for [1-13C]pyruvate peaks in

the prostate for the three patients evaluated in phase 1 with single–time point

spatially localized MRSI doses of 0.14,

0.28, and 0.43 ml/kg were in the range of

4.5 to 8.4, 6.5 to 10.6, and 14.3 to 25.4, respectively. Hence, although [1-13C]lactate

peaks were observed in tumor voxels

at the two lower doses, the higher levels

of [1-13C]pyruvate at 0.43 ml/kg allowed

for a more reliable estimate of the ratio

of [1-13C]lactate/[1-13C]pyruvate. In the

following discussion on variations in this

ratio in the prostate gland, we focus on

the results from single–time point MRSI

data that were obtained at the highest

dose during phases 1 and 2 of the study.

Representative 2D MRSI data are shown

in Fig. 4. The patient had a serum PSA of

9.5 ng/ml and bilateral biopsy-proven

Gleason grade 3 + 3 prostate cancer. The

T2 images, apparent diffusion coefficient

(ADC) images, and 1

H spectra from the

staging examination demonstrated an abnormal lesion, but only on the right side

of the gland (Fig. 4). The 13C spectra

were acquired from 33 to 45 s after the

end of the injection, and the corresponding [1-13C]lactate/[1-13C]pyruvate image overlay, which highlights in color

regions with a ratio of more than 0.6,

Time (s)

Signal-to-noise ratio

0 20 40 60 80 0

5

10

15

20

25

0 20 40 60 80

0

5

10

15

20

25

Time (s)

Signal-to-noise ratio

Prostate cancer voxel Vascular voxel

k pyr lac = 0.07 s-1

Pyruvate

Lactate

A B

k pyr lac = 0.005 s-1

Time (s)

Signal-to-noise ratio

0 20 40 60 80 0

5

10

15

20

25

Contra-lateral prostate voxel

k pyr lac = 0.016 s-1

C D

Fig. 3. 2D 13C dynamic MRSI data. Images are from a representative patient with a current PSA of

3.6 ng/ml, who had biopsy-proven prostate cancer in the left apex (Gleason grade 3 + 4) and received the

highest dose of hyperpolarized [1-13C]pyruvate (0.43 ml/kg). (A) A focus of mild hypointensity can be seen

on the T2-weighted image, which was consistent with the biopsy findings. (B to D) 2D localized dynamic

hyperpolarized [1-13C]pyruvate and [1-13C]lactate from spectral data that were acquired every 5 s from

voxels overlapping the contralateral region of prostate (turquoise), a region of prostate cancer (yellow), and a

vessel outside the prostate (green). The dynamic data were fit as described previously (22).

RESEARCH ARTICLE

www.ScienceTranslationalMedicine.org 14 August 2013 Vol 5 Issue 198 198ra108 4

by guest on March 5, 2018 http://stm.sciencemag.org/ Downloaded from

极T代谢磁共振全球科研集锦

74

demonstrated bilateral areas of hyperpolarized [1-13C]lactate. The

median [1-13C]lactate/[1-13C]pyruvate ratios were 0.94 (range, 0.85

to 1.25) on the right side and 1.35 (range, 0.73 to 4.39) on the left side of

the gland. The median [1-13C]lactate/[1-13C]pyruvate ratio in a region

of the central gland—which was not thought to include tumor—was

0.45 (range, 0.31 to 0.49). The median SNRs of [1-13C]lactate in the

right and left regions of suspected tumor were 7.6 and 6.7, with the

corresponding median SNRs of [1-13C]pyruvate being 7.8 and 5.4, respectively. MR-guided biopsy that was performed subsequent to the 13C study confirmed the presence of bilateral cancer corresponding to

these abnormalities, with Gleason grade 3 + 4 on the right and Gleason

grade 3 + 3 with high-grade prostatic intraepithelial neoplasia on the

left side of the gland. The fact that the hyperpolarized imaging method

was able to detect bilateral cancer, whereas conventional anatomic imaging methods were only able to visualize unilateral cancer, is an

exciting finding, which may be especially important in monitoring patients like this who are thought to have slow-growing cancers and are

being followed with active surveillance before starting treatment.

One concern in interpreting the data from the patient in Fig. 4 was

that the SNR of the [1-13C]pyruvate peak was relatively low in some

voxels, making it difficult to obtain an accurate estimate of the ratio of

[1-13C]lactate/[1-13C]pyruvate for tumor versus normal tissue. For this

reason, subsequent data sets were obtained starting at the earlier time

of 25 s after the end of injection rather than at 33 s. For data acquired

in this time window, peaks of [1-13C]pyruvate were observed in most

of the gland, and the criteria used to define regions of suspected tumor

within the prostate gland were an SNR of [1-13C]pyruvate >10.0, a

clearly visualized peak corresponding to [1-13C]lactate peak with an

SNR >3.0, and the ratio of the peak heights of 0.2 or higher. An example

of 3D localized 13C MRSI data acquired and evaluated in this manner is

seen in Fig. 5. The results from the corresponding MR staging examination are shown in fig. S1. This patient (patient A in Table 2) had serum

PSA of 4.5 ng/ml and a diagnosis of bilateral biopsy-proven Gleason

grade 3 + 3 cancer. The 13C spectral data showed large peaks corresponding to [1-13C]pyruvate throughout the spectral grid with lower,

but still clearly detected, [1-13C]lactate peaks in the highlighted

region. The median estimates for levels

of [1-13C]pyruvate, [1-13C]lactate, and

[1-13C]lactate/[1-13C]pyruvate for tumor

voxels identified on multiple slices as being

tumor were 41.0, 13.6, and 0.28, respectively (Table 2). The color metabolite overlays highlight regions with [1-13C]lactate/

[1-13C]pyruvate that were higher than the

cutoff value on multiple slices. The patient

had a follow-up MR-guided biopsy that

demonstrated that the lesion had progressed to Gleason 3 + 4 prostate cancer.

Results from three other patients

who had areas of elevated [1-13C]lactate/

[1-13C]pyruvate are shown in Fig. 6 and

Table 2. The same acquisition timing was

used to obtain the MRSI data and the same

cutoff criteria applied to define voxels with

abnormal metabolite ratios as for patient

A. For two of the cases (patients B and D),

bilateral regions of suspected tumor were

identified, and for the third case, there

was a unilateral focus of suspected tumor.

As can be seen in Table 2, the median

ratios of [1-13C]lactate/[1-13C]pyruvate

ranged from 0.28 to 0.38, the median SNR

for [1-13C]lactate ranged from 4.8 to 8.4,

and the median SNR for [1-13C]pyruvate

ranged from 13.5 to 30.0. The locations

highlighted in color on the metabolite

overlay images were consistent either with

the original biopsies from these patients

or with abnormalities that were identified in the images from their MR staging

examinations.

DISCUSSION

First-in-man metabolic imaging with hyperpolarized [1-13C]pyruvate was successfully

Cho Cit Cho Cit

Lac Pyr Lac

Pyr

T2 image + Lac/Pyr

ADC image

Axial T2 image T2 image + RH spectral grid T2 image + LH spectral grid

H RH array

13 13

H LH array

C RH array C LH array

1 1

Fig. 4. 2D single–time point MRSI data. Images were obtained from a patient with serum PSA of

9.5 ng/ml, who was diagnosed with bilateral biopsy-proven Gleason grade 3 + 3 prostate cancer

and received the highest dose of hyperpolarized [1-13C]pyruvate (0.43 ml/kg). The axial T2-weighted

image shows a unilateral region of reduced signal intensity (red arrows), which is consistent with a

reduction in the corresponding ADC. The 1

H spectral arrays supported these findings, with voxels

with reduced citrate and elevated choline/citrate (highlighted in pink) on the right side of the gland

and voxels with normal metabolite ratios on the left side. The 13C spectral arrays show voxels with

elevated levels of hyperpolarized [1-13C]lactate/[1-13C]pyruvate (highlighted in pink) on both the

right and left sides of the prostate. The location of colored regions in the metabolite image overlay

had a ratio of [1-13C]lactate/[1-13C]pyruvate greater than or equal to 0.6.

RESEARCH ARTICLE

www.ScienceTranslationalMedicine.org 14 August 2013 Vol 5 Issue 198 198ra108 5

by guest on March 5, 2018 http://stm.sciencemag.org/ Downloaded from

极T代谢磁共振全球科研集锦

75

Axial T2 image

T2 image + Lac/Pyr

Axial T2 image

Lac

Pyr

13C

Fig. 5. 3D single–time point localized MRSI data. Images were obtained

from patient A in Table 2, who had a serum PSA of 4.5 ng/ml, was originally

diagnosed with bilateral biopsy-proven Gleason grade 3 + 3 prostate

cancer, and received the highest dose of hyperpolarized [1-13C]pyruvate

(0.43 ml/kg). The upper panel shows an axial T2-weighted images and

corresponding spectral array with the area of putative tumor highlighted

by pink shading. A region of tumor was observed on the T2-weighted images

(red arrows), as well as on ADC images and 1

H MRSI data (fig. S1). A region

of relatively high hyperpolarized [1-13C]lactate was observed in the same

location as the abnormalities that had been observed on the multiparametric 1

H staging exam. The lower panels show axial T2 images with and without

metabolite overlays for different axial slices from the same patient. The

colored regions in these overlays have a ratio of [1-13C]lactate/[1-13C]pyruvate

≥0.2. These demonstrated a large volume of bilateral cancer.

Table 2. Estimates of metabolite parameters within regions of tumor. The values shown are the median (range) of the SNR for hyperpolarized

[1-13C]lactate, hyperpolarized [1-13C]pyruvate, and the ratios of these metabolites in tumor voxels for the patients shown in Figs. 5 and 6.

Subject n voxels [1-13C]Lactate/[1-13C]pyruvate [1-13C]Lactate SNR [1-13C]Pyruvate SNR

Patient A 11 0.28 (0.26–0.36) 13.6 (5.9–18.6) 41.0 (17.5–54.0)

Patient B 6 0.38 (0.32–0.47) 4.8 (3.6–7.8) 13.5 (10.1–20.0)

Patient C 3 0.32 (0.27–0.33) 6.3 (4.1–7.3) 22.5 (12.8–23.0)

Patient D 4 0.28 (0.24–0.33) 8.4 (6.4–16.1) 30.0 (27.8–49.4)

RESEARCH ARTICLE

www.ScienceTranslationalMedicine.org 14 August 2013 Vol 5 Issue 198 198ra108 6

by guest on March 5, 2018 http://stm.sciencemag.org/ Downloaded from

极T代谢磁共振全球科研集锦

76

applied in 31 patients with prostate cancer. There were no DLTs observed,

and the highest dose, namely, 230 mM [1-13C]pyruvate (0.43 ml/kg), was

selected for evaluation of different MR data acquisition strategies in

phase 2 of the study. The prototype polarizer and QC procedures used

to prepare the hyperpolarized agent were complex, and the fact that 31

of 33 preparations were successful in producing material that was injected into the patients is encouraging. Future studies would benefit from

the development of more robust and automated procedures for generating and preparing the hyperpolarized agent for human use. An improved

system that can simultaneously polarize four samples and uses a disposable fluid path has been designed (23) and tested in preclinical studies

(24). It is anticipated that clinical studies will take advantage of such technology to provide higher polarizations and faster agent delivery.

The kinetics of delivering the hyperpolarized agent and the increase in

[1-13C]lactate signal were consistent with those observed in preclinical

models (25–29). The 1D dynamic MRSI data obtained in our study

showed higher [1-13C]lactate signal in slices including tumor, but

also contained some signal from tissues outside of the prostate. The

[1-13C]lactate signal was low or undetectable in slices from regions of

the prostate that did not include tumor. This agrees with previous patient prostate biopsy studies, which demonstrated very low lactate

concentrations (30), and with preclinical studies, which demonstrated

a low flux of hyperpolarized [1-13C]pyruvate to lactate in normal prostate (17). The 2D dynamic MRSI data were able to distinguish signals

from tumor and vessels, with the rate of [1-13C]pyruvate to [1-13C]lactate

conversion being four to five times higher in tumor. This is consistent

with the increased expression of MCT1, MCT4, and LDH-A, and

increased activity of LDH in the tumor that was observed in preclinical studies (17, 18).

The 2D and 3D single–time point MRSI data had excellent SNRs

and spectral quality. These data sets were acquired in 8 to 12 s and

accurately reflected the presence, location, and size of cancer relative to surrounding healthy prostate tissues. Moving

to the earlier spectral acquisition window

of 25 s after injection, for the patients

shown in Figs. 5 and 6, yielded higher hyperpolarized [1-13C]pyruvate signals and

more readily quantifiable [1-13C]lactate/

[1-13C]pyruvate ratios (Table 2) than for

the patient shown in Fig. 4. The marked

changes in hyperpolarized [1-13C]pyruvate

signal with time and differences in kinetics

suggest that the acquisition of spatially localized dynamic MR data is a promising

approach for comparing metabolic parameters for tumor versus normal tissue.

Patients who participated in phase 2

of the study and who had regions of abnormal signal on their previous multiparametric MR screening examinations also had

elevated [1-13C]lactate/[1-13C]pyruvate

relative to signals from normal prostate.

In Fig. 4, our hyperpolarized imaging method highlighted a region of biopsy-proven

tumor that was not observed during the

staging examination; this is important

because it shows the power of using more

advanced technologies for characterizing tissues that appear normal

on conventional anatomic images. There were two other subjects

with findings of abnormal [1-13C]lactate/[1-13C]pyruvate in regions

where the anatomic imaging was uncertain, but in those cases, no direct biopsy evidence was available to confirm that they corresponded

to tumor.

There are several limitations of our study because it is first in human. Although 13C data acquisition parameters had been investigated

in phantom and preclinical studies, it was not possible to choose the

most appropriate values without having results from human subjects.

This led to adjustments in the timing parameters and spatial resolution being made during the course of the study. Although this did not

influence the goals of phase 1 of the study, which were to establish the

safety and feasibility of using hyperpolarized [1-13C]pyruvate in humans, it did mean that the more detailed analysis of data was restricted

to subjects participating in phase 2. Another limitation was that the

population chosen for the study had relatively early-stage cancer, and

it was not always possible to make exact correlations between the imaging findings and pathologic findings. The choice of patients with

early-stage disease who were on active surveillance was driven by

the need to avoid complications in data interpretation associated with

previous treatment and to have plenty of time to recruit and monitor

them during a period when they were expected to have stable disease.

Evidence from preclinical studies has shown that the magnitude of

[1-13C]lactate/[1-13C]pyruvate increases with tumor grade (18), and

so we believe that our technology will be even more sensitive for

evaluating patients with advanced and aggressive cancers.

When combined with findings from preclinical studies, the results

of this first-in-man study suggest that hyperpolarized 13C metabolic

imaging may be valuable for initial diagnosis and for monitoring therapy. In addition to improved technology for generating and delivering

Patient B Patient C Patient D

Fig. 6. Further representative examples of 3D single–time point MRSI data. The axial T2-weighted

images and overlays of hyperpolarized [1-13C]lactate/[1-13C]pyruvate are from the three patients labeled

as B to D in Table 2. All three of the patients had biopsy-proven Gleason grade 3 + 3 prostate cancer and

received the highest dose of hyperpolarized [1-13C]pyruvate (0.43 ml/kg). Patient B had a current PSA of

5.1 ng/ml, patient C had a PSA of 9.8 ng/ml, and patient D had a PSA of 1.9 ng/ml. The SNR and metabolite

ratios in the regions highlighted in color on the image overlays are given in Table 2.

RESEARCH ARTICLE

www.ScienceTranslationalMedicine.org 14 August 2013 Vol 5 Issue 198 198ra108 7

by guest on March 5, 2018 http://stm.sciencemag.org/ Downloaded from

极T代谢磁共振全球科研集锦

77

hyperpolarized agents becoming available (23), designing new MR

data acquisition sequences that use compressed sensing (31), parallel

imaging strategies, and multichannel 13C radio frequency coils (32, 33)

will help in translating these methods into the clinic. These developments will allow similar methods to be applied to more diverse populations of cancer patients who are undergoing surgical resection or

image-directed biopsy. In addition to having the potential for more

accurate diagnosis and staging, there have been a large number of

studies using hyperpolarized agents in preclinical models of lymphoma

(20), prostate (16–18), brain (26, 34), liver (25), breast (29), and bone

cancers (19). Other disease models that have been studied and show

promise for translation to humans are cardiac disease (33, 35, 36)

and arthritis (37).

MATERIALS AND METHODS

Patient population

Eligible patients were recruited from the prostate cancer clinic at University of California, San Francisco (UCSF), and had untreated, biopsyproven localized prostate cancer with no significant history of cardiac

or pulmonary disease and adequate baseline organ function. They

were required to have received a multiparametric 1

H MR staging examination before the hyperpolarized MR study to confirm that they

had observable lesions and ensure that they were familiar with standard imaging procedures (Supplementary Methods). The process for

generating the agent and the clinical protocol had received approval as

an Investigational New Drug (IND) from the U.S. Food and Drug

Administration. Patients provided written informed consent for participation in the imaging study.

Study design

The purpose of this study was to establish the safety and feasibility of

using hyperpolarized metabolic imaging with [1-13C]pyruvate in patients with prostate cancer. For phase 1 of the study, the commonly

used 3 + 3 dose escalation clinical trial design was modified to enroll

six patients at each dose: three to monitor the kinetics of delivery and

three to evaluate the spatial distribution of metabolism in tumor versus other tissues. Three dose levels were considered (0.14, 0.28, or

0.43 ml/kg actual body weight of 230 mM pyruvate solution). Patients

underwent continuous electrocardiogram monitoring during and for

10 min after the injection, as well as at baseline, 1 hour, and 2 hours

after the injection. Clinical monitoring was undertaken for 2 hours

after injection, with clinical and laboratory assessments performed

at 24 hours and 7 days.

Toxicities were graded with CTCAE v4.0. criteria (21). A DLT

was defined as an event of grade 2, 3, or 4 that was attributable

to the agent and occurred within 7 days after the imaging examination. The stopping rule in each cohort of six subjects was as

follows: if 0 or 1 dose DLT were observed, the study would proceed

to the next dose level; if two DLTs were observed, that dose would

become the maximum tolerated dose; and if more than two DLTs

occur, the next lower dose would be considered as the maximum

tolerated dose. Once either the maximum dose or the maximum tolerated dose cohorts were completed, phase 2 of the study design included an expansion cohort to optimize the imaging protocol and

explore the biological variability in delivery, transport, and metabolism of the agent.

Formulation

The formulation comprised [1-13C]pyruvate (22.0 mg/ml) (Supplementary Methods), sodium (4.1 mg/ml), tris (12.1 mg/ml), EDTA (0.1 mg/ml),

and tris{8-carboxyl-2,2,6,6-tetra[2-(1-methoxyethyl)]benzo(1,2-d:4,5-

d′)bis(1,3)dithiole-4-yl}methyl (radical) (4.6 mg/ml) in sterile water

for injection. All components were manufactured according to current

Good Manufacturing Practices and were provided in sterile single-use

packaging by GE Healthcare.

13C MR examination protocol

An intravenous catheter was placed, and the patient was positioned in

a clinical 3-T MR scanner (GE Healthcare). For anatomic imaging, the 1

H body coil was used for transmission with a pelvic phased array and

custom-designed 1

H/13C endorectal coil for reception (Fig. 1). For 13C

data, a bore-insertable volume coil that was hinged like a clamshell to

facilitate patient entry was used for transmission with the 13C channel

of the endorectal coil for reception (38). Anatomic images comprising

a scout, sagittal, and axial T2-weighted fast spin echo sequences were

acquired first, followed by 13C signal calibration that used the signal

from a sealed standard housed within the endorectal coil containing

600 ml of 8 M 13C-urea. Once the appropriate scan parameters had

been defined, the operators in the clean room started the dissolution.

If the sample satisfied the tests imposed by the QC system and the

pharmacist who was monitoring the study gave their approval, the

formulation was injected into the patient at ~5 ml/s and 13C data were

obtained.

Acquisition parameters for 13C MRSI data

Spatially localized dynamic 13C spectroscopic imaging monitored delivery, transport, and metabolism of hyperpolarized [1-13C]pyruvate.

Single–time point 2D or 3D acquisitions were used to obtain arrays

of 13C spectra from the prostate and surrounding tissues in 8 to 12 s.

The initial sequence parameters were chosen on the basis of studies in

murine and dog models (16, 17, 22), but were further refined as the

study progressed in terms of the start times for acquiring MR data and

the flip angle schemes used (see below).

1D dynamic MRSI data. Spectra were obtained every 3 s starting

at the end of the injection from a 36- to 60-mm axial slice encompassing the prostate. Echo planar localization was applied in the right-left

direction at 10-mm resolution and with echo time (TE)/repetition

time (TR)/flip angle of 2 ms/3 s/10°.

2D localized dynamic MRSI data. Spectra were obtained every

5 s starting at 5 s after the end of injection from a 12- to 40-mm axial

slice with eight-phase encodes in one dimension and with 18-step

echo planar localization in the other dimension to provide 10-mm

in-plane resolution. The radio frequency pulses that were used applied

either a 10° flip angle for all metabolites or a specially designed pulse

that gave a flip angle of 10° for pyruvate and 20° for lactate. The

TE/TR was 3/125 ms.

2D single–time point MRSI data. Spectra were obtained from a

10- to 20-mm axial slice with 12 by 12–phase encodes and 7-mm inplane resolution with a progressive flip angle and TE/TR of 3/85 ms.

The acquisition time was 12 s and started 25 to 33 s after the end of

the injection. The variation in slice thickness was required to cover the

region of putative tumor.

3D single–time point MRSI data. A 3D array of spectra was obtained from a 43- to 120-mm axial slice with 12 by 8 to 12–phase encodes and 18 echo planar frequency encodes. The in-plane resolution

RESEARCH ARTICLE

www.ScienceTranslationalMedicine.org 14 August 2013 Vol 5 Issue 198 198ra108 8

by guest on March 5, 2018 http://stm.sciencemag.org/ Downloaded from

极T代谢磁共振全球科研集锦

78

was 7 mm and the through-plane resolution was 7 to 15 mm, with a

progressive flip angle and TE/TR of 3/85 to 125 ms. Variations of the

in-plane slice thickness, number of phase encodes, and spatial resolution were driven by differences in the size of the prostate and the spatial

extent of the region of tumor. The acquisition time was 8 to 12 s, and

it started 25 to 33 s after the end of the injection.

Data analysis

The analysis of the 13C MR data used specialized software developed

in our laboratory (39). Arrays of spectra were obtained by apodizing

the raw data with a 10-Hz Lorentzian function in the time domain

and performing a Fourier transform. For data with echo planar localization, signals from the positive and negative gradient lobes were separately reconstructed for each trajectory with regridding of samples on

the ramps. The spectral arrays were then zero- and first-order phasecorrected. Quantification of individual spectra used automatic phasing,

baseline subtraction, and frequency correction. The heights and areas

of spectral peaks were estimated and used to generate metabolite images and/or curves of the time course of changes in [1-13C]lactate and

[1-13C]pyruvate.

The spectral arrays and metabolite images were directly correlated

with anatomic images that were acquired within the same examination. For comparison purposes, regions of prostate cancer were identified as areas with concordant positive TRUS-guided biopsy and MRI

abnormality within the same sextant of the prostate. Visual comparisons of the locations of regions with elevated lactate/pyruvate on the

13C images were made with the results from the MR staging examination using anatomic images from the two studies as a reference to see

whether similar regions were identified as having abnormalities. For

voxels where the SNR of the dynamic data was sufficient, the curves

of lactate and pyruvate were fit with the two-compartment model that

was developed and applied in previous preclinical studies (22).

SUPPLEMENTARY MATERIALS

www.sciencetranslationalmedicine.org/cgi/content/full/5/198/198ra108/DC1

Methods

Fig. S1. Results from the MR staging examination for a patient with a large volume of bilateral

cancer.

Table S1. Summary of information about individual patients.

Table S2. Number of adverse events that were observed and their grade as defined by criteria

from the National Cancer Institute.

REFERENCES AND NOTES

1. R. Siegel, D. Naishadham, A. Jemal, Cancer statistics, 2012. CA Cancer J. Clin. 62, 10–29

(2012).

2. M. Han, A. W. Partin, S. Piantadosi, J. I. Epstein, P. C. Walsh, Era specific biochemical recurrencefree survival following radical prostatectomy for clinically localized prostate cancer. J. Urol.

166, 416–419 (2001).

3. J. E. McNeal, D. G. Bostwick, R. A. Kindrachuk, E. A. Redwine, F. S. Freiha, T. A. Stamey,

Patterns of progression in prostate cancer. Lancet 1, 60–63 (1986).

4. M. R. Cooperberg, P. R. Carroll, L. Klotz, Active surveillance for prostate cancer: Progress

and promise. J. Clin. Oncol. 29, 3669–3676 (2011).

5. J. T. Wei, R. L. Dunn, H. M. Sandler, P. W. McLaughlin, J. E. Montie, M. S. Litwin, L. Nyquist,

M. G. Sanda, Comprehensive comparison of health-related quality of life after contemporary

therapies for localized prostate cancer. J. Clin. Oncol. 20, 557–566 (2002).

6. J. E. Johansson, O. Andrén, S. O. Andersson, P. W. Dickman, L. Holmberg, A. Magnuson,

H. O. Adami, Natural history of early, localized prostate cancer. JAMA 291, 2713–2719

(2004).

7. A. Bill-Axelson, L. Holmberg, M. Ruutu, M. Häggman, S. O. Andersson, S. Bratell, A. Spångberg,

C. Busch, S. Nordling, H. Garmo, J. Palmgren, H. O. Adami, B. J. Norlén, J. E. Johansson;

Scandinavian Prostate Cancer Group Study No. 4, Radical prostatectomy versus watchful

waiting in early prostate cancer. N. Engl. J. Med. 352, 1977–1984 (2005).

8. S. S. Mehta, D. P. Lubeck, N. Sadetsky, D. J. Pasta, P. R. Carroll, Patterns of secondary cancer

treatment for biochemical failure following radical prostatectomy: Data from CaPSURE.

J. Urol. 171, 215–219 (2004).

9. G. J. Kelloff, P. Choyke, D. S. Coffey; Prostate Cancer Imaging Working Group, Challenges in

clinical prostate cancer: Role of imaging. AJR Am. J. Roentgenol. 192, 1455–1470 (2009).

10. J. Kurhanewicz, D. B. Vigneron, Advances in MR spectroscopy of the prostate. Magn. Reson.

Imaging Clin. N. Am. 16, 697–710 (2008).

11. J. C. Weinreb, J. D. Blume, F. V. Coakley, T. M. Wheeler, J. B. Cormack, C. K. Sotto, H. Cho,

A. Kawashima, C. M. Tempany-Afdhal, K. J. Macura, M. Rosen, S. R. Gerst, J. Kurhanewicz,

Prostate cancer: Sextant localization at MR imaging and MR spectroscopic imaging

before prostatectomy—Results of ACRIN prospective multi-institutional clinicopathologic

study. Radiology 251, 122–133 (2009).

12. J. M. Beauregard, S. G. Williams, T. R. Degrado, P. Roselt, R. J. Hicks, Pilot comparison of

F-fluorocholine and F-fluorodeoxyglucose PET/CT with conventional imaging in prostate

cancer. J. Med. Imaging Radiat. Oncol. 54, 325–332 (2010).

13. J. H. Ardenkjaer-Larsen, B. Fridlund, A. Gram, G. Hansson, L. Hansson, M. H. Lerche, R. Servin,

M. Thaning, K. Golman. Increase in signal-to-noise ratio of >10,000 times in liquid-state NMR.

Proc. Natl. Acad. Sci. U.S.A. 100, 10158–10163 (2003).

14. K. Golman, R. in ‘t Zandt, M. Thaning, Real-time metabolic imaging. Proc. Natl. Acad. Sci. U.S.A.

103, 11270–11275 (2006).

15. K. Golman, J. S. Petersson, Metabolic imaging and other applications of hyperpolarized 13C1

. Acad. Radiol. 13, 932–942 (2006).

16. A. P. Chen, M. J. Albers, C. H. Cunningham, S. J. Kohler, Y. F. Yen, R. E. Hurd, J. Tropp, R. Bok,

J. M. Pauly, S. J. Nelson, J. Kurhanewicz, D. B. Vigneron, Hyperpolarized C-13 spectroscopic

imaging of the TRAMP mouse at 3T—Initial experience. Magn. Reson. Med. 58, 1099–1106

(2007).

17. K. R. Keshari, R. Sriram, M. Van Criekinge, D. M. Wilson, Z. J. Wang, D. B. Vigneron, D. M. Peehl,

J. Kurhanewicz, Metabolic reprogramming and validation of hyperpolarized 13C lactate as a

prostate cancer biomarker using a human prostate tissue slice culture bioreactor. Prostate

73, 1171–1181 (2013).

18. M. J. Albers, R. Bok, A. P. Chen, C. H. Cunningham, M. L. Zierhut, V. Y. Zhang, S. J. Kohler,

J. Tropp, R. E. Hurd, Y. F. Yen, S. J. Nelson, D. B. Vigneron, J. Kurhanewicz, Hyperpolarized

13C lactate, pyruvate, and alanine: Noninvasive biomarkers for prostate cancer detection and

grading. Cancer Res. 68, 8607–8615 (2008).

19. H. Dafni, P. E. Larson, S. Hu, H. A. Yoshihara, C. S. Ward, H. S. Venkatesh, C. Wang, X. Zhang,

D. B. Vigneron, S. M. Ronen, Hyperpolarized 13C spectroscopic imaging informs on hypoxiainducible factor-1 and myc activity downstream of platelet-derived growth factor receptor.

Cancer Res. 70, 7400–7410 (2010).

20. S. E. Day, M. I. Kettunen, F. A. Gallagher, D. E. Hu, M. Lerche, J. Wolber, K. Golman,

J. H. Ardenkjaer-Larsen, K. M. Brindle, Detecting tumor response to treatment using

hyperpolarized 13C magnetic resonance imaging and spectroscopy. Nat. Med. 13,

1382–1387 (2007).

21. National Cancer Institute, Common Terminology Criteria for Adverse Events (CTCAE) v4.0;

http://ctep.cancer.gov.

22. M. L. Zierhut, Y. F. Yen, A. P. Chen, R. Bok, M. J. Albers, V. Zhang, J. Tropp, I. Park,

D. B. Vigneron, J. Kurhanewicz, R. E. Hurd, S. J. Nelson, Kinetic modeling of hyperpolarized

13C1-pyruvate metabolism in normal rats and TRAMP mice. J. Magn. Reson. 202, 85–92

(2010).

23. J. H. Ardenkjaer-Larsen, A. M. Leach, N. Clarke, J. Urbahn, D. Anderson, T. W. Skloss,

Dynamic nuclear polarization polarizer for sterile use intent. NMR Biomed. 24, 927–932

(2011).

24. S. Hu, P. E. Larson, M. Vancriekinge, A. M. Leach, I. Park, C. Leon, J. Zhou, P. J. Shin, G. Reed,

P. Keselman, C. von Morze, H. Yoshihara, R. A. Bok, S. J. Nelson, J. Kurhanewicz, D. B. Vigneron,

Rapid sequential injections of hyperpolarized [1-13C]pyruvate in vivo using a sub-kelvin, multisample DNP polarizer. Magn. Reson. Imaging 31, 490–496 (2013).

25. S. Hu, A. Balakrishnan, R. A. Bok, B. Anderton, P. E. Larson, S. J. Nelson, J. Kurhanewicz,

D. B. Vigneron, A. Goga, 13C-pyruvate imaging reveals alterations in glycolysis that precede

c-Myc-induced tumor formation and regression. Cell Metab. 14, 131–142 (2011).

26. I. Park, R. Bok, T. Ozawa, J. J. Phillips, C. D. James, D. B. Vigneron, S. M. Ronen, S. J. Nelson,

Detection of early response to temozolomide treatment in brain tumors using hyperpolarized

13C MR metabolic imaging. J. Magn. Reson. Imaging 33, 1284–1290 (2011).

27. B. Pullinger, H. Profka, J. H. Ardenkjaer-Larsen, N. N. Kuzma, S. Kadlecek, R. R. Rizi, Metabolism

of hyperpolarized [1-13C]pyruvate in the isolated perfused rat lung—An ischemia study.

NMR Biomed. 25, 1113–1118 (2012).

28. C. S. Ward, H. S. Venkatesh, M. M. Chaumeil, A. H. Brandes, M. Vancriekinge, H. Dafni,

S. Sukumar, S. J. Nelson, D. B. Vigneron, J. Kurhanewicz, C. D. James, D. A. Haas-Kogan,

S. M. Ronen, Noninvasive detection of target modulation following phosphatidylinositol

3-kinase inhibition using hyperpolarized 13C magnetic resonance spectroscopy. Cancer Res.

70, 1296–1305 (2010).

RESEARCH ARTICLE

www.ScienceTranslationalMedicine.org 14 August 2013 Vol 5 Issue 198 198ra108 9

by guest on March 5, 2018 http://stm.sciencemag.org/ Downloaded from

极T代谢磁共振全球科研集锦

79

29. T. H. Witney, M. I. Kettunen, D. E. Hu, F. A. Gallagher, S. E. Bohndiek, R. Napolitano, K. M. Brindle,

Detecting treatment response in a model of human breast adenocarcinoma using hyperpolarised [1-13C]pyruvate and [1,4-13C2]fumarate. Br. J. Cancer 103, 1400–1406 (2010).

30. M. B. Tessem, M. G. Swanson, K. R. Keshari, M. J. Albers, D. Joun, Z. L. Tabatabai, J. P. Simko,

K. Shinohara, S. J. Nelson, D. B. Vigneron, I. S. Gribbestad, J. Kurhanewicz, Evaluation of

lactate and alanine as metabolic biomarkers of prostate cancer using 1H HR-MAS

spectroscopy of biopsy tissues. Magn. Reson. Med. 60, 510–516 (2008).

31. P. E. Larson, S. Hu, M. Lustig, A. B. Kerr, S. J. Nelson, J. Kurhanewicz, J. M. Pauly, D. B. Vigneron,

Fast dynamic 3D MR spectroscopic imaging with compressed sensing and multiband

excitation pulses for hyperpolarized 13C studies. Magn. Reson. Med. 65, 610–619 (2012).

32. A. Arunachalam, D. Whitt, K. Fish, R. Giaquinto, J. Piel, R. Watkins, I. Hancu, Accelerated

spectroscopic imaging of hyperpolarized C-13 pyruvate using SENSE parallel imaging.

NMR Biomed. 22, 867–873 (2009).

33. A. Z. Lau, A. P. Chen, N. R. Ghugre, V. Ramanan, W. W. Lam, K. A. Connelly, G. A. Wright,

C. H. Cunningham, Rapid multislice imaging of hyperpolarized 13C pyruvate and

bicarbonate in the heart. Magn. Reson. Med. 64, 1323–1331 (2010).

34. M. Marjańska, I. Iltis, A. A. Shestov, D. K. Deelchand, C. Nelson, K. Uğurbil, P. G. Henry, In vivo 13C spectroscopy in the rat brain using hyperpolarized [1-13C]pyruvate and [2-13C]pyruvate.

J. Magn. Reson. 206, 210–218 (2010).

35. K. X. Moreno, S. M. Sabelhaus, M. E. Merritt, A. D. Sherry, C. R. Malloy, Competition of pyruvate

with physiological substrates for oxidation by the heart: Implications for studies with hyperpolarized [1-13C]pyruvate. Am. J. Physiol. Heart Circ. Physiol. 298, H1556–H1564 (2010).

36. M. A. Schroeder, L. E. Cochlin, L. C. Heather, K. Clarke, G. K. Radda, D. J. Tyler, In vivo assessment of pyruvate dehydrogenase flux in the heart using hyperpolarized carbon-13

magnetic resonance. Proc. Natl. Acad. Sci. U.S.A. 105, 12051–12056 (2008).

37. J. D. MacKenzie, Y. F. Yen, D. Mayer, J. S. Tropp, R. E. Hurd, D. M. Spielman, Detection of

inflammatory arthritis by using hyperpolarized 13C-pyruvate with MR imaging and

spectroscopy. Radiology 259, 414–420 (2011).

38. J. Tropp, P. Calderon, D. Vigneron, Systems, methods and apparatus for an endo-rectal

receive-only probe, U.S. Patent 7945308 (2011).

39. J. C. Crane, M. P. Olson, S. J. Nelson, SIVIC: An open-source DICOM software framework and

plug-in for OsiriX PACS, International Society for Magnetic Resonance in Medicine 18th

Scientific Meeting, 2010.

Acknowledgments: We acknowledge the important contributions to the success of the trial

from M. McPolin, B. Jimenez, S. Hu, J. Crane, H. Yoshihara, C. Sotto, K. Grycz, and M. Contreras

from UCSF; C. Cunningham from the University of Toronto; and P. Calderon, T. Skloss, P. Sontum,

M. Thaning, G. Torheim, J. Wolber, Y.-F. Yen, and M. Mendizabal from GE Healthcare. Funding:

Pulse sequence and clinical study design were supported by NIH grants R01 EB007588 and R21

EB005363. The polarizer and costs of the 13C patient studies were supported with funding from GE

Healthcare. Author contributions: S.J.N., D.B.V., and J.K. planned the project and coordinated and

oversaw the patient studies. P.E.Z.L. developed pulse sequences and acquired and processed the

13C data. A.L.H. was the oncologist who designed the patient studies, obtained the IND, and cared

for patients. M.F. was the pharmacist responsible for all aspects of the compounding process. M.v.C.

was responsible for engineering design and polarizer function. J.W.C. was responsible for the clean

room, polarizations, and dissolutions. R.B. monitored patients. I.P. was responsible for postprocessing

and data analysis. G.R. acquired preclinical and patient data. L.C. was responsible for specialized

hardware and radio frequency coils. E.J.S. and P.M. participated in designing the patient studies.

V.K.W. provided statistical input for study design and interpretation of the results. J.H.A.-L. was responsible for the polarizer design and for providing advice on protocol development. A.P.C. and

R.E.H. contributed to engineering development. L.-I.O. coordinated the provision of sterile components and the material used to produce the imaging agent. F.J.R. provided input on radio

frequency coil design and system integration. J.T. designed, constructed, and tested the radio

frequency coils. J.A.M. developed the study concept and coordinated the academic-industry

partnership that made the clinical trial possible. Competing interests: Some of the MRI acquisition

methods that were used have been patented (#7,795,868). The authors have no other competing

interests to declare. Data and materials availability: Information about the pulse sequence design

and data processing methods is available through the NIH-supported Hyperpolarized MRI Technology

Resource Center at UCSF (www.radiology.ucsf.edu/research/labs/hyperpolarized-mri-tech). The

endorectal coil is patented by General Electric and cannot be made and sold for profit except

under license.

Submitted 5 March 2013

Accepted 19 July 2013

Published 14 August 2013

10.1126/scitranslmed.3006070

Citation: S. J. Nelson, J. Kurhanewicz, D. B. Vigneron, P. E. Z. Larson, A. L. Harzstark, M. Ferrone,

M. van Criekinge, J. W. Chang, R. Bok, I. Park, G. Reed, L. Carvajal, E. J. Small, P. Munster,

V. K. Weinberg, J. H. Ardenkjaer-Larsen, A. P. Chen, R. E. Hurd, L.-I. Odegardstuen, F. J. Robb,

J. Tropp, J. A. Murray, Metabolic imaging of patients with prostate cancer using hyperpolarized

[1-13C]pyruvate. Sci. Transl. Med. 5, 198ra108 (2013).

RESEARCH ARTICLE

www.ScienceTranslationalMedicine.org 14 August 2013 Vol 5 Issue 198 198ra108 10

by guest on March 5, 2018 http://stm.sciencemag.org/ Downloaded from

极T代谢磁共振全球科研集锦

80

Hyperpolarized 1-[13C]-Pyruvate Magnetic

Resonance Imaging Detects an Early Metabolic Response to Androgen Ablation Therapy in Prostate Cancer

研究背景和研究过程

研究结果

研究对象

ג极ࣅ) HP) 13C 磁共振࠼೷ׯ) ၟMRSI) ๟ᅃዖႎ႙ׯၟरຍLj੗ᅜ੺໏࿮ںظ֪॔ۯༀཚୟ༬ᅴႠ代谢ࢅ૙ิײࡗLj

ժሞ 13C Քऻڦ࿿ิݴጱׯၟڦଳ௺܈ฉइڥമ໯࿄ᆶڦሺᅮDŽሺे 10000-200000 ԠDž๟ాᇸႠڦĂ࿮ࢅڦ۾࿮ݣ

พႠڦăፕኁኮമԒڢକዷۯ֪॔ڦമଚ၇ҵ࣒ኁዐ HP[13C]- եཛྷ໗ MRSI ڦْ๯ට༹ 1 ೺ଣض研৯Ljժኤํକሞߛ

प՚ዐվइे໏ HP եཛྷ໗ - ළ໗ཚଉ൶ᇘڦ੗ႜႠăᇑگप՚ҵኢᇑଆႠፇኯă

ሞኄ૛Ljፕኁ௮ຎକڼᅃ߲঳ࡕLjኤ௽କ๑ᆩ HP [13C]- եཛྷ໗ MRSI ܔႧघ໎Ԋܡଐ݆ (ADT) ڦ代谢ݒᆌ

代՗Ⴀዡၠ T2 े඄ (T2W) ঴೪཮ၟࢅ၎ᆌڦ՗຤࠵ક

ොဣຕ (ADC) ཮ၟࢅ T2W ཮ၟᇑ۠ेڦեཛྷ໗ - ළ໗

代谢ཚଉ (kPL) ཮ၟࢅ၎ᆌגڦ极ࣅ) HP) 13C ࠼೷ንଚ

(SA) ᅃఁ 52 ໠മଚ၇ҵ࣒ኁLjሞዎଐമDŽADžࢅDŽBDž

ሞਸ๔Ⴇघ໎ၩබࣅࢅଐࢫ 6 ዜ࣒ᆶߛڦݘ࠽प՚മ

ଚ၇ҵăሞዎଐമLj੗ᅜൣںؤੂڟമଚ၇ҵ൶ᇘDŽࢤ

෥ॱཀྵDžLjᅺྺ T2W ࢅ ADC ཮ၟฉڦ႑ࡽ३ณLjHP

ළ໗ࢅ HP 13C MRI ฉ၎ڦ࠲ kPL ཚଉሺेă ਸ๔Ⴇघ໎

ԊܡዎଐࢫLjHP ළ໗ࢅ kPL ၂ጣইگ዁ኟ׉ೝ຤Ljܔ

മଚ၇༹ओࢅ ADC ৈᆶ๢ڦ܈ዎଐၳࡕă

ᅃఁമଚ၇ҵ࣒ኁ

极T代谢磁共振全球科研集锦

81

结 论

应用方向

T2 े඄ MRI ฉ዗ୀۇٴၭڦՎࣅ੗ᅜࢮ୼փऺLjADC ׯၟኻᆶ๢܈ՎࣅLjኄኧ׼ HP13C MRI ሞ๑ᆩՔጚݣพბՔጚඓ

ۨኄዖݒᆌኮമॠ֪ራ೺代谢ݒᆌڦ૰ీăᇑኄၵ݀၄ᅃዂLj࣒ኁໜࢫइڥକ၂ጣڦଣݒضᆌLjሞ ADT ਸ๔ࢫ 6 ߲

ሆ ้ეൣ PSA ፌۅگ݆࿮ॠ֪ڟă

ڼᅃ߲࣒ኁ๖૩ຫ௽କ HP[13C]- եཛྷ໗ׯၟፕྺݒᆌ代谢ิ࿿Քኾ࿿ڦയ૰ă৊ᅃօڦଣض研৯ֱۙ HP13C MRI ڦ代

谢Վࣅᇑዎଐݒᆌڸࢅੇኮक़࠲ڦဣኟሞ৊ႜዐă

മଚ၇ҵ

极T代谢磁共振全球科研集锦

82

Research Letter

Hyperpolarized 1-[13C]-Pyruvate Magnetic Resonance Imaging

Detects an Early Metabolic Response to Androgen Ablation

Therapy in Prostate Cancer

Rahul Aggarwal *, Daniel B. Vigneron, John Kurhanewicz

Hyperpolarized (HP) 13C magnetic resonance spectroscopic

imaging (MRSI) is a novel imaging technique that allows

rapid and noninvasive monitoring of dynamic pathwayspecific metabolic and physiologic processes [1] with

unprecedented gain in sensitivity (10 000–200 000 fold

increase) for imaging of 13C-labeled biomolecules that are

endogenous, nontoxic, and nonradioactive [2,3]. We previously reported the first-in-human phase 1 clinical study of

HP [

13C]-pyruvate MRSI in patients with prostate cancer on

active surveillance, and confirmed the feasibility of

capturing regions of accelerated HP pyruvate-to-lactate

flux in high-grade versus low-grade cancer versus benign

tissue [4].

Here we describe the first results demonstrating the

metabolic response to androgen deprivation therapy (ADT)

using HP [

13C]-pyruvate MRSI. The patient presented with

serum prostate-specific antigen (PSA) of 25.2 ng/ml and

Gleason 4 + 5 prostate adenocarcinoma on biopsy.

Cross-sectional imaging demonstrated metastases within

the pelvic nodes and osseous structures. Baseline multiparametric (mp) 1

H MRI of the prostate (anatomic

imaging, diffusion-weighted imaging [DWI], dynamic

E U R O P E A N U R O L O G Y X X X ( 2 0 17 ) X X X – X X X

available a t www.sciencedirect.com

journal homepage: www.europeanurology.com

Fig. 1 – Representative axial T2-weighted (T2W) anatomic image and corresponding water apparent diffusion coefficient (ADC) image and T2W image

with an overlaid pyruvate-to-lactate metabolic flux (kPL) image and corresponding hyperpolarized (HP) 13C spectral array (SA) for a 52-yr-old prostate

cancer patient with extensive high-grade prostate cancer (A) before therapy and (B) 6 wk after initiation of androgen ablation and chemotherapy.

Before treatment, the region of prostate cancer can be clearly seen (red arrows) as a reduction in signal on the T2W and ADC images, and increased HP

lactate and associated kPL flux on HP 13C MRI. After initiation of androgen deprivation therapy there was a significant reduction in reduction in HP

lactate and kPL to normal levels, with only a modest treatment effect on prostate volume and ADC.

EURURO-7484; No. of Pages 2

Please cite this article in press as: Aggarwal R, et al. Hyperpolarized 1-[13C]-Pyruvate Magnetic Resonance Imaging Detects an

Early Metabolic Response to Androgen Ablation Therapy in Prostate Cancer. Eur Urol (2017), http://dx.doi.org/10.1016/j.

eururo.2017.07.022

http://dx.doi.org/10.1016/j.eururo.2017.07.022

0302-2838/© 2017 Published by Elsevier B.V. on behalf of European Association of Urology.

极T代谢磁共振全球科研集锦

83

contrast-enhanced [DCE] imaging, and 3D 1H MRSI) with HP

[

13C]-pyruvate revealed a bulky tumor involving the left

apex, mid gland, and base peripheral and transition zones,

and right apex, mid gland, and base peripheral zone,

measuring 4.5 ! 1.5 ! 5.1 cm3

. T2-weighted MRI showed a

well-defined focus of low signal intensity (T2 score 5/5;

Fig. 1A). The lesion also had marked restricted diffusion

(DWI score 5/5; apparent diffusion coefficient [ADC] 930)

and was DCE-positive, with increased uptake and washout

of contrast agent, and MRSI-positive, with elevated choline

and reduced citrate on 1H MRSI. The overall Prostate

Imaging-Reporting and Data System v.2 score was 5.

Figure 1A shows the HP 13C spectral array for the baseline

scan, with markedly elevated lactate peaks within tumorcontaining voxels. A color scale map of dynamic pyruvateto-lactate metabolic flux (kPL) values likewise shows

markedly elevated flux levels in the tumor compared to

adjacent normal tissue in the baseline HP [

13C]-pyruvate

MRI.

At 6 wk after initiation of ADT, repeat imaging

demonstrated nearly complete abrogation of elevated HP

lactate peaks on HP 13C MRI (Fig. 1B) and associated near

complete diminution of intratumoral kPL values on dynamic

imaging (kPL max 0.025 s"1 at baseline and 0.007 s"1 on

follow-up). Notably, there was negligible change in size of

tumor on T2-weighted MRI and only a modest change on

ADC imaging, supporting the ability of HP 13C MRI to detect

early metabolic responses before such a response can be

ascertained using standard radiographic criteria. Concordant with these findings, the patient subsequently achieved

a marked clinical response, with an undetectable serum PSA

nadir at 6 mo after ADT initiation.

This first patient example illustrates the potential of HP

[

13C]-pyruvate imaging as a metabolic biomarker of

response. Further clinical studies investigating the association between metabolic changes on HP 13C MRI and

response and resistance to treatment are ongoing.

Conflicts of interest: The authors have nothing to disclose.

Acknowledgments: This work was supported by NIH grants

R01EB017449, P41EB013598, and R01CA166655. We would like to

acknowledge the following for their contribution: Robert A. Bok, Peder E.

Z. Larson, Jeremy W. Gordon, Hsin-Yu Chen, Marcus Ferrone, James B.

Slater, Mark van Criekinge, Lucas Carvajal, Sarah J. Nelson, Eric J. Small,

Matt Cooperberg, Pamela N. Munster, and Albert Chang.

References

[1] Chen AP, Kurhanewicz J, Bok R, et al. Feasibility of using hyperpolarized [1-13C]lactate as a substrate for in vivo metabolic 13C MRSI

studies. Magnetic Resonance Imaging 2008;26:721–6.

[2] Kurhanewicz J, Vigneron DB, Brindle K, et al. Analysis of cancer

metabolism by imaging hyperpolarized nuclei: prospects for translation to clinical research. Neoplasia 2011;13:81–97.

[3] Ardenkjaer-Larsen JH, Fridlund B, Gram A, et al. Increase in signalto-noise ratio of 10,000 times in liquid-state NMR. Proc Natl Acad

Sci U S A 2003;100:10158–63.

[4] Nelson SJ, Kurhanewicz J, Vigneron DB, et al. Metabolic imaging of

patients with prostate cancer using hyperpolarized [1-13C]pyruvate. Sci Transl Med 2013;14:198ra108.

University of California San Francisco, San Francisco, CA, USA

*Corresponding author. University of California San Francisco,

1600 Divisadero Street, San Francisco, CA 94131, USA.

Tel. +1 415 3539278; Fax: +1 415 3537779.

E-mail address: rahul.aggarwal@ucsf.edu (R. Aggarwal).

July 19, 2017

2 E U R O P E A N U R O L O G Y X X X ( 2 0 17 ) X X X – X X X

EURURO-7484; No. of Pages 2

Please cite this article in press as: Aggarwal R, et al. Hyperpolarized 1-[13C]-Pyruvate Magnetic Resonance Imaging Detects an

Early Metabolic Response to Androgen Ablation Therapy in Prostate Cancer. Eur Urol (2017), http://dx.doi.org/10.1016/j.

eururo.2017.07.022

极T代谢磁共振全球科研集锦

84

Investigation of analysis methods for hyperpolarized 13C-pyruvate metabolic MRI in

prostate cancer patients

研究背景

研究对象

研究过程

ଣض๬ᄓዐኟሞ研৯๑ᆩג极ࣅ) HP) ༐ 13 եཛྷ໗ڦ MRILjᅜ༵ࠃҵኢࢅ႐ሤׯၟ代谢ݥڦ෇ൔႠ֪ଉăԨၜణዐLj

ፕኁॽ HP[1-13C] եཛྷ໗ۯༀ MRI ᆌᆩᇀമଚ၇ҵዐLjᅜ֪ଉեཛྷ໗ڟළ໗ڦገࣅLjᇨऺሞൔလႠҵኢዐገࣷࣅሺेă

Ԩ研৯ڦణՔ๟ਸ݀ࢅ๬֪ݴဆݛ݆Ljᅜ߀৊ኄዖ代谢ገڦࣅଉࣅă

17ఁമଚ၇ҵ࣒ኁ

ਏᆶג极ࣅ) HP)13C Քऻڹ࿿ڦ MRI ᅙׯྺᅃዖ׉ݥᆶമ཰ڦ代谢ׯၟݛ๕Ljᅺྺ໲੗ᅜ༑֪࣒ኁ研৯ዐ࠲ڦ॰代谢

཰০ă ኄၵ研৯૧ᆩම঴ۯༀࢃ极ࣅ) DNP) ઠሺഽ 13C ࢃ极ࣅLjీࠕሞ༹ాሺे > 50000 Ԡڦଳ௺܈ă ങৃྺኹLjፌ

ᆶထྭڦ HP ڹ] ๟࿿1-13C] եཛྷ໗Lj໲༵ࠃᅜူࠀీǖ֪॔࠲॰代谢཰০Ljገࣅ] ྺ1-13C] ළ໗Ljࢫኁሞܠۇٴຕҵኢ

ૌ႙ዐ܈ߛฉۙDŽĐWarburg ၳᆌđDž; ཚࡗම঴ DNP ํ၄܈ߛ极ࣅǗ ሞ ڼ1 ೺Ҿ全๬ᄓዐඓۨڦवଉྺ 0.43 ࡹ/ื

ࠅ༹ুዘࢅ 250 mM ้ڦ࿿ิ၎ඹႠă

HP13C MRI ዐ代谢ገڦࣅଉࣅ๟޿ఇ๕ଣضᆌᆩ࠲ڦ॰ፇׯևݴă ໜጣ HP13C MRI ৊෇ڦݘ࠽ଣض๬ᄓLjጚඓĂ࿘ॳ

ዘᄲă࠲዁݆ݛଉ֪ڦᆶᅪᅭ൐

փ࠶ླྀጀڦֶᅴසࢆLjHP13C MRI ዐۯڦༀׯ֑ၟ集༵ࠃକܔ代谢ገࣅ৊ႜ࿘ॳଉڦࣅയ૰ă

Ԩ研৯ኼሞೠࠚᆩᇀටૌ HP13C MRI研৯ڦ代谢ଉݛࣅ݆ăྺକํ၄ኄᅃۅLjፕኁೠࠚକएᇀටૌമଚ၇ҵۯༀ MR

࠼೷ׯ) ၟMRSI) 研৯ዐ࠵ִڦڟ༬ኙڦఇెݛ݆ă ፕኁ࣏ॽኄၵݛ݆ᆌᆩᇀമଚ၇ҵ࣒ኁ HP13C եཛྷ໗ຕ਍ڦా༹

ݴဆLj၂๖ڦڟڥ kPL ኵᅜतഄ໱ํᄓ༬ኙLj૩ස႑ሯԲ(SNR)ࢅླྀጀ֖ຕă

极T代谢磁共振全球科研集锦

85

研究结果

结 论

应用方向

代谢 AUC ڦ཮ా༹ᄣᅜत໯ᆶెࢇ֖ຕă T2 മଚ၇թՎ

ᆯ T2 े඄঴೪֖੊཮ၟዐڦ෥୴ॱཀྵ཭؜၂๖ă tSNR ࢅ

kPL ཮ၟྺ௅߲ዷ༶܀૬ेشă ෙ߲ kPL ཮ၟሞ௅߲ዷ༶ዐ

شڦ੨၎ཞă ໯ᆶ้क़ኵDŽೝ਩้क़ pyrĂT1LĂTarrival ࢅ

TbolusDžۼᅜ s ྺڇ࿋Ljժ൐ሞ๴๬ኁኮक़֑ᆩ၎ཞشڦ੨ă

ెࢇ֖ຕৈሞ tSNRpyruvate > 80 ้ऺ໙ࢅ၂๖Ljᅜ༵ࠃ੗

ጀ༬ླྀڦ T1P ᅺࣷࢇె෇๼ᇑ ă܈৛ߛ༵ྲ୼Ljࢇెڦ੍

ႠՎࣅ๴ܸڟೋֶă ໯ᆶݛ݆ۼ٪ሞ T1L Վࢅࣅ RF ݀พࠀ

୲ဃֶڦೋֶă

Ԩ研৯ኤ௽ג极ࣅ༐ 13 եཛྷ໗ڦ MRI ీࠕ๑ᆩۯༀׯၟۯࢅ૰ბ

ఇ႙༵ࠃമଚ၇ҵ代谢ڦۨଉೠࠚLjժ༵؜മଚ၇ዐླྀጀڿໃĂ

ݛბॺఇ૰ۯ研৯全ևएᇀਏᆶԨăྷݔݔࡀڦ୲ࣅ代谢ገࢅ SNR

ࣅăᆩᇀଉऺࠚ代谢ڦ࠲࿮ໃ༬Ⴀڿጀླྀᇑࠃ༵Ljᅜၟׯༀۯڦ݆

代谢ڦ AUC Բ୲ݛ݆ሞ้࢛ۨक़݋ገঙڦཉॲူLjᅜतሞླྀጀߴ

ᄱኮമਸ๔֑集ຕ਍้๟࿘ॳڦLjڍ ๴T1L ੗ՎႠڦᆖၚă࿮๼෇

kPL ెݛࢇ݆ܔᇀ໯ᆶ݋ገঙݛӄࢅླྀጀ༬Ⴀگڦ SNR ຕ਍၂๖၎

ܔ࿘ॳLjڍᄺܔ T1L ڦ੗ՎႠ௺ߌăੵ研৯ڦླྀጀٳڟ֪ଉኵ٪ሞ

ࡗג 10 ௱ڦֶᅴLjമଚ၇ాጺ SNR ڦֶᅴᆶबԠLj࿄ઠ研৯ዐ๑

ᆩڦ֑集ݴࢅဆݛ݆ՂႷ๢ᆌኄዖֶᅴă

മଚ၇ҵ

极T代谢磁共振全球科研集锦

86

Received: 4 April 2018 Revised: 14 June 2018 Accepted: 28 June 2018

DOI: 10.1002/nbm.3997

RESEARCH ARTICLE

Investigation of analysis methods for hyperpolarized

13C-pyruvate metabolic MRI in prostate cancer patients

Peder E. Z. Larson1,2 Hsin-Yu Chen1,2 Jeremy W. Gordon1 Natalie Korn1,2

John Maidens3 Murat Arcak3 Shuyu Tang1,2 Mark Criekinge1 Lucas Carvajal1

Daniele Mammoli1 Robert Bok1 Rahul Aggarwal4 Marcus Ferrone5 James B. Slater1

Sarah J. Nelson1,2 John Kurhanewicz1,2 Daniel B. Vigneron1,2

1Department of Radiology and Biomedical

Imaging, University of California – San

Francisco, San Francisco, California

2UC Berkeley–UCSF Graduate Program in

Bioengineering, University of California,

Berkeley and University of California, San

Francisco, California

3Department of Electrical Engineering and

Computer Sciences, University of

California – Berkeley, Berkeley, California

4Department of Medicine, University of

California – San Francisco, San Francisco,

California

5Department of Clinical Pharmacy, University

of California – San Francisco, San Francisco,

California

Correspondence

Peder E. Z. Larson, Department of Radiology

and Biomedical Imaging, University of

California – San Francisco, San Francisco,

California.

Email: peder.larson@ucsf.edu

Funding information

National Institutes of Health, Grant/Award

Number: R01EB017449, R01EB016741,

R01CA183071, R01CA211150 and

P41EB013598

MRI using hyperpolarized (HP) carbon-13 pyruvate is being investigated in clinical trials to provide non-invasive measurements of metabolism for cancer and cardiac imaging. In this project,

we applied HP [1-13C]pyruvate dynamic MRI in prostate cancer to measure the conversion from

pyruvate to lactate, which is expected to increase in aggressive cancers. The goal of this work was

to develop and test analysis methods for improved quantification of this metabolic conversion.

In this work, we compared specialized kinetic modeling methods to estimate the

pyruvate-to-lactate conversion rate, kPL, as well as the lactate-to-pyruvate area-under-curve

(AUC) ratio. The kinetic modeling included an “inputless” method requiring no assumptions

regarding the input function, as well as a method incorporating bolus characteristics in the fitting. These were first evaluated with simulated data designed to match human prostate data,

where we examined the expected sensitivity of metabolism quantification to variations in kPL,

signal-to-noise ratio (SNR), bolus characteristics, relaxation rates, and B1 variability. They were

then applied to 17 prostate cancer patient datasets.

The simulations indicated that the inputless method with fixed relaxation rates provided high

expected accuracy with no sensitivity to bolus characteristics. TheAUC ratio showed an undesired

strong sensitivity to bolus variations. Fitting the input function as well did not improve accuracy

over the inputless method. In vivo results showed qualitatively accurate kPL maps with inputless

fitting. The AUC ratio was sensitive to bolus delivery variations. Fitting with the input function

showed high variability in parameter maps.

Overall, we found the inputless kPL fitting method to be a simple, robust approach for quantification of metabolic conversion following HP [1-13C]pyruvate injection in human prostate cancer

studies. This study also provided initial ranges of HP [1-13C]pyruvate parameters (SNR, kPL, bolus

characteristics) in the human prostate.

KEYWORDS

hyperpolarized MRI, kinetic modeling, metabolic imaging, prostate cancer, 13C-pyruvate

1 INTRODUCTION

MRI with hyperpolarized (HP) 13C-labeled substrates has emerged as an extremely promising metabolic imaging modality, because it can probe key

metabolic pathways in patient studies.1-6 These studies utilize dissolution dynamic nuclear polarization (DNP) to enhance the 13C nuclear polarization, providing > 50 000 fold sensitivity increases in vivo.7,8 The most promising HP substrate to date is [1-13C]pyruvate, which provides the

Abbreviations used: AUC, area-under-curve; DCE, dynamic contrast-enhanced; DNP, dynamic nuclear polarization; EPSI, echo-planar spectroscopic imaging; HP, hyperpolarized; MRSI, MR

spectroscopic imaging; SNR, signal-to-noise ratio; tSNR, total SNR; TTP, time-to-peak; VIF, vascular input function

NMR in Biomedicine. 2018;e3997. wileyonlinelibrary.com/journal/nbm © 2018 John Wiley & Sons, Ltd. 1 of 17

https://doi.org/10.1002/nbm.3997

极T代谢磁共振全球科研集锦

87

2 of 17 LARSON ET AL.

following: monitoring of a key metabolic pathway, conversion to [1-13C]lactate, which is highly upregulated in most cancer types (the “Warburg

effect”); high levels of polarization via dissolution DNP; and biocompatibility at doses of 0.43 ml/kg body weight and 250mM as determined in a

Phase 1 safety trial.1 Clinical translation of HP [1-13C]pyruvate was initially demonstrated in prostate cancer patients, and currently there is clinical

research underway at several institutions in prostate cancer, breast cancer, brain tumors, liver metastases, and heart failure.1-5

Quantification of metabolic conversion in HP 13C MRI is a key component for the clinical application of this modality. Accurate, robust, and meaningful measurement methods are essential as HP 13C MRI enters widespread clinical trials. The robustness of the acquisition and quantification

methods will be especially critical when comparing data between institutions and in multisite clinical trials. The preclinical studies performed to justify clinical studies benefited from highly reproducible experimental and physiological conditions, whereas in humans substantially more variability

is a concern.

Dynamic imaging acquisition in HP 13C MRI offers the potential to provide robust quantification of metabolic conversion, regardless of differences in bolus delivery.9,10 This is in contrast to imaging at a single time-point, which can be analyzed via the lactate-to-pyruvate ratio,11 but is very

dependent on experimental timing. Dynamic imaging acquisitions can be analyzed using kinetic modeling, typically to calculate a pyruvate-to-lactate

metabolic conversion rate, kPL.

12-23 Another popular approach is to use the area-under-curve (AUC) ratio between lactate and pyruvate that, under

assumptions of constant-in-time flip angles acquired starting prior to bolus arrival or consistent bolus characteristics, is directly proportional to

kPL.

24

The purpose of this work was to evaluate methods for quantification of metabolism for human HP 13C MRI studies. To accomplish this, we evaluated methods in simulations that were based on the observed characteristics in human prostate cancer dynamic MR spectroscopic imaging (MRSI)

studies. We also applied these methods for in vivo analysis of prostate cancer patient HP 13C-pyruvate data. The resulting kPL values are presented,

along with additional experimental characteristics such as signal-to-noise ratio (SNR) and bolus parameters.

2 METHODS

2.1 Dynamic imaging

Dynamic hyperpolarized 13C MRSI was acquired for all patients analyzed in this article. They were imaged using a 3D dynamic MRSI sequence

that covered the entire prostate using a blipped echo-planar spectroscopic imaging (EPSI) acquisition with a compressed sensing reconstruction,25

shown in Figure 1. This sequence used a flyback EPSI waveform with 16 encodes, a spectral resolution of 9.83 and a 581-Hz spectral bandwidth. A compressed sensing reconstruction based on spectral, spatial and temporal sparsity allowed for 18-fold acceleration compared with fully

sampled EPSI.

Other 3DMRSI sequence parameters included 12 × 12 × 16 matrix size, TE=6.3 ms, TR=150 ms, 8 mm isotropic resolution, acquisition window

= 42 s, and 2 s between time points. With the accelerated acquisition, only eight encodes were required per time point. Multiband spectral–spatial

RF excitation pulses were used, with a lower flip angle applied to the pyruvate resonance in order to minimize saturation and maintain substrate

magnetization for later time points. This results in improved SNR for the metabolic products.10 This was combined with a variable flip-angle strategy

in time to use all HP magnetization.26-28 Pyruvate variable flip angles were designed with a “T1-effective” approach, using an effective decay rate

of T1,eff = 35 s in Equations 6 and 7 of Xing et al28 for the flip-angle design. Lactate variable flip angles were designed using the maximum SNR

formulation in Equation 8 of Nagashima,27 with an effective decay rate of 100 s for the flip-angle design. These effective decay rates for the flip-angle

designs were empirically chosen to maintain the SNR for both metabolites throughout the dynamic acquisition and to provide some robustness to

B1 inhomogeneity based on simulations. The flip angles used are shown in Figure 1.

The pulses were designed using minimal spectral specifications, where flip angles for [1-13C]pyruvate and [1-13C]lactate were specified by the

methods described above, with all other resonances designated as “don't-care” regions. This allowed for a relatively short duration of 4.3 ms, compared with 18–20 ms for previous designs that also had specifications for [1-13C]alanine and [1-13C]pyruvate-hydrate.10 The MATLAB software

used to design these pulses is available at https://github.com/LarsonLab/Spectral-Spatial-RF-Pulse-Design.29

2.2 Data analysis

The primary goal of the kinetic model fitting methods was to evaluate the pyruvate-to-lactate conversion rates for our human prostate cancer imaging studies. These used the imaging methods described above. The data typically had relatively low SNR and several key experimental parameters

were unknown: the in vivo metabolite relaxation rates, bolus delivery shape, and timing.

We used the tissue model shown in Figure 2, which includes an input function outside the imaging voxel and unidirectional pyruvate-to-lactate

conversion via kPL within the voxel:

u → P k P L−−→L (1)

which can be described by the differential equations

dPZ(t)

dt = −R1PPZ(t) − kPLPZ(t) + u(t) (2)

dLZ(t)

dt = −R1LLZ(t) + kPLPZ(t) (3)

极T代谢磁共振全球科研集锦

88

LARSON ET AL. 3 of 17

FIGURE 1 Dynamic MRSI acquisition scheme. Data were encoded with a blipped EPSI acquisition and all time points were reconstructed together

using compressed sensing.25 Excitation was performed using multiband spectral-spatial RF pulses, which applied different flip angles for pyruvate

and lactate. The flip angles were also increased over time, to preserve magnetization for capturing dynamics while using all HP magnetization by

the end of the experiment

In this nomenclature, PZ and LZ denote the pyruvate and lactate magnetization, u is the incoming magnetization, kPL is the pyruvate-to-lactate

conversion rate, and R1P = 1∕T1P, R1L = 1∕T1L are the spin-lattice relaxation rates.

To account for arbitrary flip-angle schemes, we used a hybrid discrete–continuous model.12,30 Flip-angle compensation is achieved by converting the measured signal for metabolite X at time point n, XS[n], to the Z magnetization prior to RF excitation, X−

Z [n]. The Z magnetization after RF

excitation, X+

Z [n], is also computed, and these conversions are based on the cumulative effects of the RF pulses required to acquire each image:

X−

Z [n] = XS[n]∕SS,X[n] (4)

X+

Z [n] = X−

Z [n]SZ,X[n] (5)

SS,X[n] = ∑Nrf

i=1

sin 휃X,n[i]

∏i−1

j=1

cos 휃X,n[j] (6)

SZ,X[n] = ∏Nrf

i=1

cos 휃X,n[i] (7)

Nrf is the number of RF pulses used to acquire each image and 휃X,n[i] are the set of flip angles for metabolite X used to acquire the time-point n image.

We also assumed that the input function, u(t), was constant over each TR interval, which enabled an analytic solution to Equations 2 and 3 during

the continuous period between RF pulses of our discrete–continuous model.

We compared the following metabolism quantification strategies in the framework of the above tissue model.

• Inputless kPL fitting.

This fitting approach, inspired by Khegai et al,17 only fits the lactate magnetization and not the pyruvate magnetization, where the measured

pyruvate magnetization is used as the input for the kinetic model at each time point.

The estimated lactate magnetization measurement ̂

L−

Z [n] at each time point, n, is fit based only on the measured pyruvate magnetization at the

adjacent time points, P+

Z [n − 1], P−

Z [n], and the estimated lactate magnetization at the previous time point ̂

L+

Z [n − 1] by solving the two-site model

in differential Equations 2 and 3:

极T代谢磁共振全球科研集锦

89

4 of 17 LARSON ET AL.

FIGURE 2 Tissue model and resulting example magnetization curves. The tissue model included an input function outside the imaging voxel, with

a unidirectional input. Within the voxel, this model includes unidirectional pyruvate-to-lactate conversion via kPL and loss of magnetization due to

RF pulses and relaxation. Example longitudinal and transverse magnetization curves using this model are shown, including a gamma input function,

using the nominal simulation parameters of kPL = .02 /s, T1P = 30 s, T1L = 25 s, Tarrival = 4 s, Tbolus = 12 s, TR = 2 s, and the in vivo multiband variable

flip-angle strategy

[

P

̂ −

Z [n]

̂

L−

Z [n]

]

= x⋆[n − 1] + exp(A · TR)

([ P

̂ +

Z [n − 1]

̂

L+

Z [n − 1]

]

− x⋆[n − 1]

)

(8)

where

A =

[

−R1P − kPL 0

kPL −R1L

]

and

x⋆[n − 1] = A−1

[

u[n − 1]

0

]

This solution assumes a constant input, u[n − 1], during each time interval, TR, between time points n − 1 and n, which is computed based on the

measured pyruvate magnetization:

极T代谢磁共振全球科研集锦

90

LARSON ET AL. 5 of 17

u[n − 1] =

(P−

Z [n] − P+Z [n − 1] exp((−R1P − kPL)TR))(R1P + kPL)

1 − exp((−R1P − kPL)TR) . (9)

Themodel was solved based onminimization of the least-squares error as computed between themeasured and estimated lactatemagnetization,

Σn(L−

Z [n] − ̂

L−

Z [n])2, using a constrained, nonlinear least-squares solver based on a trust-region-reflective algorithm (MATLAB).

• Area-under-curve ratio24 (AUCratio).

For this method, the ratio of the area under lactate to the area under pyruvate curves is used as a simple surrogate for metabolic conversion

and is computed simply as

AUCratio =

∑

nLS[n]

∑

nPS[n] (10)

Under conditions of sampling prior to arrival of the bolus and constant-in-time flip angles (i.e. not variable flip angles), this ratio is24

AUCratio ≈ kPL

R1L,eff

(11)

where R1L,eff is the effective lactate relaxation rate, including T1 decay, flow, conversion from lactate to pyruvate, and losses due to RF pulses.

For the multiband variable flip scheme, we computed a “calibrated AUCratio”, where predicted AUC values were computed from simulated data

without noise using the nominal model parameters, including the nominal input function.

• kPL fitting with input.

In this fitting approach, the input function, u(t), was included in the fitting process and then pyruvate and lactate magnetization were fitted.

This was done using the discrete–continuous model described above to include varying flip angles. A gamma function was used for the input

function, u(t):

u(t) = { 0, t < Tarrival;

ku(t − Tarrival)

훼 exp(−(t − Tarrival)∕훽), t ≥ Tarrival

(12)

with 훼 = 4 and 훽 = Tbolus∕4 to provide the shape shown in Figure 2.

This model had additional parameters of Tarrival, Tbolus, and ku (an input rate), which could be either fitted or fixed. The shape of the bolus can

also be modified readily in this approach.

The model was solved based on minimization of the least-squares error as computed between the measured and estimated pyruvate and

lactate magnetization, Σn(L−

Z [n] − ̂

L−

Z [n])2+ (P−Z [n] − P

̂ −

Z [n])2, using a constrained, nonlinear least-squares solver based on a trust-region-reflective

algorithm (MATLAB).

• Lactate time-to-peak (TTP) and mean lactate time.

Recent work by Daniels et al14 showed that the lactate timing changes with kPL and the lactate time-to-peak (TTP) model-free approach performed indistinguishably from the best kinetic model in both their in vitro and in vivo datasets. Simulation results of the metric, as well as a ‘mean

lactate time’, are described in the Supporting Information.

Constraints were placed on several parameters in the kPL fitting methods. For all results shown, these were T1L,min = 15 s, T1L,max = 35 s,

Tarrival,min = 0 s, Tarrival,max = 12 s, Tbolus,min = 6 s, Tbolus,max = 10 s.

To assess the variability in arrival time between the experiments, we computed a “mean pyruvate time” as the center of mass of the pyruvate

signal over time31:

T휇,pyr =

∑

nPS[n]t[n]

∑

nPS[n] (13)

where t[n] was the time after the nth image acquisition. This metric is reflective of the relative arrival and was chosen as it is relatively robust to

noise compared with estimates of peak times which is essential given the low SNR of our data and requires no assumptions regarding the pyruvate

dynamics.

Pyruvate and lactate signals were extracted from the spectra using peak area integration. The complex-valued spectra were phased separately

for pyruvate and lactate, with a zero-order correction that maximized the real component across all time points. Only the real component of

the integrated peak areas was used in the fitting. This phasing and peak integration provided metabolite signals with zero-mean Gaussian noise,

unlike magnitude-based methods, which result in Rician noise. For Gaussian noise, least-squares minimization is equivalent to maximum-likelihood

estimation.

The total SNR, tSNR, was measured for pyruvate and lactate as the summed signal divided by the standard deviation in the peak height

measurements, 휎, which was estimated using voxels outside the sensitive region of the RF coil that contained only noise:

tSNRX =

∑

nXS[n]

휎 (14)

极T代谢磁共振全球科研集锦

91

6 of 17 LARSON ET AL.

The fitting code used in this article was implemented in MATLAB and is available in the “Hyperpolarized MRI Toolbox” at https://github.com/

LarsonLab/hyperpolarized-mri-toolbox.32

2.3 Simulations

Simulations provide a key tool for evaluation of the proposed analysis methods, as there are no available gold-standard in vivo kPL measurements.

Simulated data were generated based on a two-site model as shown above, with a gamma-distribution input function and using the same flip-angle

schemes as the in vivo studies. These data were fitted as described above. Monte Carlo simulations were performed by adding Gaussian random

noise to evaluate the precision and accuracy of the fitting methods. These were also performed across ranges of values for kPL, SNR, T1L, T1P, arrival

time, and input function width, where the ranges were chosen based on what we observed for the human prostate cancer studies.

The nominal simulation values were kPL = .02 /s, T1P = 30 s, T1L = 25 s. The simulations were normalized to have a total input magnetization

of 1, and the nominal experiment had a noise standard deviation 휎 = 0.004. This noise level can be interpreted by noting that the theoretical

maximum SNR would be 250 if all HP magnetization were captured in a single image. The gamma-distribution function was nominally set to have

an arrival time 4 s after the beginning of the experiment (Tarrival = 4 s) and a full width at half-maximum of 12 s (Tbolus = 12 s), illustrated in Figure 2.

The gamma-distribution function used a shape parameter 훼 = 4 and a scaling parameter 훽 = Tbolus∕4, which was found empirically to provide

realistic bolus shapes. ForAUCratio measurements, predicted AUC values were computed from simulated data without noise using the nominal model

parameters, including the nominal input function.

Code for generating the simulations is available at https://github.com/agentmess/Prostate-Cancer-Analysis-Methods-Paper-2018.

2.4 Experiments

Imaging was performed on a GE 3TMR system on software version DV25 equipped with broad-band capabilities. For 13C, the “clamshell” transmitter

consisted of a Helmholtz pair built into the patient table was used for volume excitation and was large enough to fit the entire pelvis and other coils.

An endo-rectal probe containing both 13C and 1H loop coils33 was used for reception. For anatomical imaging, an additional four-channel 1H torso

coil was used for reception, and the 1H body coil was used for transmit. The endo-rectal probe contained a 13C-urea (8 M, 0.6 mL) syringe in the

middle of the 13C loop that was used for calibration. It was doped with gadolinium (Gd) to reduce the relaxation times to T1 = 1.0 s, T2 = 195 ms.

The T1 shortening allowed for more rapid coil testing and B1 calibration prior to the pyruvate injection.

All 17 patients (63±8 years old) used in this study had biopsy proven prostate cancer and received a multiparametric 1H MRI/hyperpolarized 13C

MR exam prior to definitive treatment for their cancer. 12 patients, five with low grade (Gleason ≤ 3+4, serum PSA 5.2±2.2 ng/ml) and seven with

high grade (Gleason ≥ 4+3, serum PSA 7.1±1.8 ng/ml), were enrolled in a clinical trial of hyperpolarized [1-13C]pyruvate MR prior to surgery with

whole-mount step section pathology (NCT02526360). The other cohort, six patients with advanced prostate cancer (Gleason Score ≥ 4+3, serum

PSA 16±0.8 ng/ml), was enrolled in a clinical trial of hyperpolarized [1-13C]pyruvate MR prior to and after therapy with androgen deprivation

therapy (NCT02911467).

Dissolution DNP was performed using a 5T SPINLab (GE Healthcare).7 The injected solution contained 241±10mM hyperpolarized

[1-13C]pyruvate polarized to 39.6±4.5% and was administered at doses of 0.43 mL/kg. In the dissolution process, the 15mM electron paramagnetic

agent (EPA), AH111501 (GE Healthcare), required for DNP was filtered out with mechanical filtration. An automated quality-control system evaluated pH, temperature, polarization, EPA and pyruvate concentration, and sample volume prior to injection. The resulting pH was 7.3±0.4 and had

an EPA concentration of 1.0±0.4 휇M.

In all experiments, the dynamic imaging acquisition was started 5 s after completion of the saline flush that followed the HP pyruvate injection.

To improve the B0 homogeneity, we used the same shimming procedure as for our 1H MRSI clinical prostate studies.34 This begins with automatic

shimming over the prostate (as selected by a PRESS selection region), followed by manual adjustments of the shim gradients to produce the peak

water signal in the selection region.

For analysis, fitting was only performed in voxels with a minimum pyruvate tSNR > 80 (Equation 14). When using fixed relaxation rates in the

fitting, we set T1P = 30, T1L = 25 s based on fits from the Phase I prostate cancer patient trial (T1P = 29.2±5, T1L = 25.2±5 s).1 Other studies in

animals have measured T1P = 43 s in whole blood extracted during the experiment and T1L = 28 to 35 s in an implanted fibrosarcoma at 7T,16 and

T1L ≈ 25 s with multiple fitting approaches in subcutaneous mammary adenocarcinomas at 3T.14

3 RESULTS

3.1 Simulations

3.1.1 Matching in vivo data

We first examined our in vivo data, and attempted to match our simulated data in order to perform relevant simulation analyses. Figure 3 shows

typical data from prostate voxels, chosen to cover a representative range of kPL, SNR, and bolus characteristics, and visually matched simulated data.

Examination of our in vivo data was used to choose our nominal simulation parameter values of kPL = .02 /s, 휎 = 0.004, Tarrival = 4 s, Tbolus = 12 s,

T1P = 30 s, T1L = 25 s, as well as the ranges of kPL, 휎, Tarrival, and Tbolus shown in subsequent analyses.

极T代谢磁共振全球科研集锦

92

LARSON ET AL. 7 of 17

FIGURE 3 Examples of human prostate hyperpolarized pyruvate data from four different patient studies and corresponding simulated data with

empirically matched parameters. All simulated data used T1P = 30 s and T1L = 25 s. Examples were chosen to span a range of parameters

3.1.2 kPL fitting with assumed bolus characteristics and relaxation rates

The simulations in Figure 4 compare the accuracy and precision of the AUCratio and kPL fitting methods using the in vivo experimental parameters

and simulated across variability ranges chosen based on the in vivo fitting results (described in greater detail later). In this initial comparison, the

fitting methods assumed several fixed parameters: the calibrated AUCratio was generated assuming known bolus characteristics and relaxation rates,

and the fitting with input method also assumed known bolus characteristics and relaxation rates that were fixed in the fitting. The simulations with

varying bolus characteristics and relaxation rates demonstrate the response to kinetic modeling with errors in these assumed parameters. The

results for this are summarized in Table 1.

极T代谢磁共振全球科研集锦

93

8 of 17 LARSON ET AL.

FIGURE 4 Sensitivity of metabolic rate estimates based on Monte Carlo simulations for fitting methods with fixed relaxation rates and bolus

characteristics. Sensitivity plots show the fractional kPL error from the kinetic models, or as predicted by an AUCratio that was calibrated for the

nominal experimental parameters (pulse sequence, bolus characteristics, and relaxation rates). These are plotted over kPL, noise level, bolus arrival

time (Tarrival), bolus duration (Tbolus), and metabolite relaxation rates. Accuracy/bias is shown by the solid lines, which are the average fit across the

simulation. Precision/variance is shown by the dashed lines, which plot ± 1 standard deviation in the simulation fits

The simulations show, in the top row in Figure 4, comparable precision and accuracy between the three methods in response to variations in kPL

and SNR. However, the other plots show that the methods differ in their response to variations in the bolus characteristics, relaxation rates, and B1.

The simulations show that the calibrated AUCratio is sensitive to errors in the bolus characteristics and relaxation rates for our prostate cancer

experimental approach. With known bolus characteristics and relaxation rates but variation in kPL and noise (top row), the precision and accuracy

of the calibrated AUCratio are comparable to those of the inputless kPL fitting. However, the AUCratio shows substantial bias with changes in all other

parameters: the bolus characteristics as well as both T1P and T1L. For example, the plots show that errors in the assumed Tarrival will result in just over

20% bias in kPL at Tarrival = 0 and Tarrival = 8. Errors in the assumed bolus duration, Tbolus, lead to up to 10% bias for the range shown. Errors of ±10 s

in the relaxation rates lead to approximately 10% bias in kPL. Errors in B1 of ±20% also lead to a similar 10% bias. Note that the sensitivity of the

calibrated AUCratio to the bolus characteristics can be eliminated by using flip-angle schemes that are constant in time and starting before the bolus

arrival, as demonstrated in the Supporting Information.

Similarly, the simulations in Figure 4 show that fitting with input that assumed known bolus characteristics and relaxation rates is also very sensitive to errors in the bolus characteristics and relaxation rates. If the bolus measurement and assumed relaxation rates are accurate, the fitting with

input is accurate for the ranges of kPL and SNR shown, with a precision slightly lower than the calibrated AUCratio. However, this approach shows

the largest bias when there are errors in the assumed bolus characteristics. With bolus arrival and duration errors of approximately 2 s, there is

极T代谢磁共振全球科研集锦

94

LARSON ET AL. 9 of 17

TABLE 1 Summary of the simulation results shown in Figures 4, 5, and 6. The values shown are the mean average error ±the standard deviation. To

highlight the weak points of each approach, values between 5 and 10% are highlighted in yellow, between 10 and 15% in orange, and > 15% in red

FIGURE 5 Sensitivity of metabolic rate estimates for fitting methods, similar to Figure 4 (fixed bolus characteristics, fixed T1P) but with fitting T1L

approximately 20% bias in kPL. It showed similar sensitivity to T1L errors to the other methods, but was more sensitive to T1P errors than the other

methods. It was relatively insensitive to errors in B1, with < 5% bias for ±20% errors.

The inputless fitting method was the most favorable compared with the other methods in Figure 4 when fixing the bolus characteristics and

relaxation rates with the chosen tissue model and experimental parameters. Unlike the other approaches, it was robust to changes in the bolus

characteristics and T1P. It was also as precise as the other methods, with similar expected variances in kPL estimates. The remaining weaknesses of

this method are bias, with variations in T1L and errors in the RF transmit power. Errors of ±10 s in T1L lead to 10–20% bias in kPL, while errors in B1

of ±20% also lead to a 20% bias.

3.1.3 kPL fitting with relaxation rate fitting

To attempt to address the bias of the fitting methods in the presence of T1L assumption errors, we also investigated inclusion of T1L fitting in Figure 5.

While this does eliminate this bias for both inputless fitting and fitting with input, adding this additional fitting parameter decreases the expected

precision of kPL estimates substantially. The expected standard deviation of kPL fitting is more than doubled in most situations shown. In response to

±10 s variations in T1L, the expected variance with T1L fitting is similar to the bias introduced by assuming a fixed T1L.

3.1.4 kPL fitting with bolus fitting

To attempt to address the sensitivity of fitting with input to the bolus characteristics, we investigated additionally fitting the bolus characteristics

instead of assuming and fixing values in Figure 6. We investigated fitting the bolus duration (Tbolus), fitting the bolus arrival time (Tarrival), and fitting

both duration and arrival time.