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

Findings suggest that ¹³C-pyruvate–to-lactate conversion can serve as an early indicator of response to multityrosine kinase inhibitor pazopanib in an ovarian cancer model; findings also suggest that evaluation of late events in glycolysis, such as by hyperpolarized ¹³C-pyruvate MR spectroscopy, may be predictive of response even when imaging of early events, such as by ¹⁸F-FDG PET, is not predictive.

Abstract

Purpose

To assess in a mouse model whether early or late components of glucose metabolism, exemplified by fluorine 18 (¹⁸F) fluorodeoxyglucose (FDG) positron emission tomography (PET) and hyperpolarized carbon 13 (¹³C)–pyruvate magnetic resonance (MR) spectroscopy, can serve as indicators of response in ovarian cancer to multityrosine kinase inhibitor pazopanib.

Materials and Methods

In this Animal Care and Use Committee approved study, 17 days after the injection of 2 × 10⁶ human ovarian SKOV3 tumors cells into 14 female nude mice, treatment with vehicle or pazopanib (2.5 mg per mouse peroral every other day) was initiated. Longitudinal T2-weighted MR imaging, dynamic MR spectroscopy of hyperpolarized pyruvate, and ¹⁸F-FDG PET/computed tomographic (CT) imaging were performed before treatment, 2 days after treatment, and 2 weeks after treatment.

Results

Pazopanib inhibited ovarian tumor growth compared with control (0.054 g ± 0.041 vs 0.223 g ± 0.112, respectively; six mice were treated with pazopanib and seven were control mice; P < .05). Significantly higher pyruvate-to-lactate conversion (lactate/pyruvate + lactate ratio) was found 2 days after treatment with pazopanib than before treatment (0.46 ± 0.07 vs 0.31 ± 0.14, respectively; P < .05; six tumors after treatment, seven tumors before treatment). This was not observed with the control group or with ¹⁸F-FDG PET/CT imaging.

Conclusion

The findings suggest that hyperpolarized ¹³C-pyruvate MR spectroscopy may serve as an early indicator of response to tyrosine kinase (angiogenesis) inhibitors such as pazopanib in ovarian cancer even when ¹⁸F-FDG PET/CT does not indicate a response.

^© RSNA, 2017

Online supplemental material is available for this article.

Introduction

Ovarian cancer is the leading cause of death from gynecologic cancers (1). Therapies that include tyrosine kinase inhibitors such as pazopanib can alter both tumor cell signaling and angiogenesis, and these can alter cell growth and metabolism. Presently, clinical response to therapeutics is generally based on decreased tumor size, but this can take months and delay decisions of whether to continue with effective therapy or stop ineffective therapy. Additional early functional information regarding response could allow early treatment modification and avoid unnecessary toxicity. Malignant lesions exhibit altered metabolism compared with normal tissues. Upstream metabolic events can be imaged by positron emission tomography (PET) with the glucose analog fluorine 18 (¹⁸F) fluorodeoxyglucose (FDG). It recently became possible to image the downstream molecule pyruvate and its metabolites by using hyperpolarized magnetic resonance (MR) imaging. Whether early or late metabolic events can serve as early predictors of response to tyrosine kinase inhibitors such as pazopanib in ovarian cancer needs further evaluation.

¹⁸F-FDG PET imaging is clinically used to help detect and stage ovarian cancer. It has been used to monitor efficacy of carboplatin and paclitaxel therapy (2) and as an early indicator of response to a vascular disrupting agent (3) in mouse ovarian cancer models. ¹⁸F-FDG enters the cell via glucose transporters and is phosphorylated by hexokinase, but this phosphorylated form is not a good substrate for phosphoglucose isomerase and its charge prevents escape from the cell. Therefore, ¹⁸F-FDG accumulation reflects early events in glycolysis. Downstream, normal glycolysis produces pyruvate, a key molecule at the branching point of entering aerobic versus anaerobic respiration. Pyruvate can be produced by other pathways such as by way of amino acid metabolism (4). Hyperpolarization can increase signal 10 000–100 000-fold (5) and, in conjunction with MR imaging, has made it possible to image exogenous hyperpolarized pyruvate and its metabolites such as lactate. Most cancers are hyperproliferative and therefore tend to have high metabolic demand.

Pazopanib inhibits multiple tyrosine kinase receptors by way of adenosine triphosphate competition for receptor phosphorylation, including c-kit, c-fms, fibroblast growth factor receptor, platelet-derived growth factor receptor, and vascular endothelial growth factor receptor (6–8) (ie, receptors involved in regulating tumor cell growth and metabolism, and receptors that regulate angiogenesis, a process critical for obtaining oxygen and nutrients for metabolism). Pazopanib showed promise as a monotherapy in patients with ovarian cancer in a phase II study (9) and extended median progression-free survival by 5.6 months in a phase III study. Metabolism can be regulated by tumor cell receptor activation and nutrient supply. Our purpose was to assess in a mouse model whether early or late components of glucose metabolism, exemplified by ¹⁸F-FDG PET and hyperpolarized carbon 13 (¹³C)–pyruvate MR spectroscopy, can serve as indicators of response in ovarian cancer to multityrosine kinase inhibitor pazopanib.

Materials and Methods

Hyperpolarized ¹³C-Pyruvate

Twenty-six milligrams of [1-¹³C]-pyruvic acid (Sigma-Aldrich) with 15 mmol/L OX063 (GE Healthcare, Fairfield, Conn) and 1.5 mmol/L Prohance (Bracco Diagnostics, Monroe Township, NJ) were hyperpolarized in a dynamic nuclear polarization system (HyperSense DNP Polarizer; Oxford Instruments, Austin, Tex) as previously described (5–10). The sample was dissolved in 4 mL of buffer, which consisted of 40 mmol/L Trizma preset crystals (pH, 7.6), 80 mmol/L sodium hydroxide, 50 mmol/L sodium chloride, and 0.1 mg/L ethylenediaminetetraacetic acid. Once dissolution was complete, 0.2 mL of 80 mmol/L hyperpolarized [1-¹³C] pyruvate (nominally 30% polarization) was drawn into a syringe for injection into the animals.

Cell Lines and Culture

The human ovarian cancer cell line SKOV3 (11) (a kind gift from Dr Anil K. Sood, Department of Gynecologic Oncology, MD Anderson Cancer Center, Houston, Tex) was maintained in Roswell Park Memorial Institute medium 1640 supplemented with 15% fetal bovine serum and 50 μg/mL gentamicin sulfate (Gemini Bioproducts, West Sacramento, Calif). Cell lines were validated by short tandem repeat DNA fingerprinting by using a short tandem repeat analysis kit (16 HS System; Promega, Madison, Wis). The short tandem repeat profiles were compared with online search databases (Deutsche Sammlung von Mikroorganismen und Zellkulturen, American Type Culture Collection, Japanese Cancer Research Resources Bank, and Rikagaku Kenkyusho) of 2455 known profiles, along with the MD Anderson–characterized cell line core (CCLC database of 2556 known profiles). The short tandem repeat profiles matched known DNA fingerprints or were unique. For subcutaneous injection, cells were dissociated by using trypsin, centrifuged at 1000 rotations per minute × 7 minutes at 4°C, washed twice, and suspended again in serum-free buffered salt solution (HBSS; Life Technologies, Rockville, Md).

In Vivo Mouse Model

Fourteen female athymic nude mice (6–8 weeks old) were purchased from the experimental radiation oncology department of the MD Anderson Cancer Center and housed in specific pathogen-free conditions. They were cared for in accordance with guidelines set forth by the American Association for Accreditation of Laboratory Animal Care and all studies were approved and supervised by the MD Anderson Cancer Center Animal Care and Use Committee.

For tumor implantation, mice were anesthetized by using isoflurane. The skin over the injection site was cleaned with 70% ethanol. We subcutaneously injected 2 × 10⁶ SKOV3 human ovarian cancer cells into the right hind leg of mice by using a syringe with a 28-gauge needle. At weekly intervals after injection, the mice were monitored for tumor growth. After 2 weeks, the mice were imaged. Then, mice were randomly assigned into two groups (seven mice per group), treated with pazopanib (2.5 mg per mouse orally every alternative day for 2 weeks) or with vehicle control. One mouse in the treated group died of unknown cause after pretherapy imaging. Tumors were measured by calipers every 2 days. PET and MR imaging was performed 1–2 days before therapy, 2 days after the initiation of therapy, and at 2 weeks after initiation of therapy. Mice were killed after final imaging and tumors were weighed.

In Vivo MR Spectroscopy

Experiments were performed by using a 7-T and 30-cm (bore size) system (BioSpec System, Bruker Biospin, Billerica, Mass) with B-GA12 gradients (120-mm inner diameter). A ¹³C-surface coil (15-mm inner diameter) was placed directly over the tumor, and the animal holder was then placed inside a dual-tuned ¹³C/hydrogen 1 (¹H)–volume coil with a 72-mm inner diameter (Bruker Biospin). The ¹H channel of this volume coil was used for ¹H MR imaging signal excitation and reception, and the ¹³C channel of the volume coil was used for ¹³C signal excitation, whereas the ¹³C surface coil was used to help detect the ¹³C signal with high sensitivity to the tumor at the center of the coil. Transverse ¹H scout images and T2-weighted images were acquired by using a rapid acquisition with relaxation enhancement spin-echo pulse sequence (repetition time msec/echo time msec, 2500/50; field of view, 50 × 50 mm; 256 × 192 matrix; two averages per increment; echo train, eight; section thickness, 1.5 mm; imaging time, 2 minutes). Dynamic ¹³C spectra were acquired by using a section-localized pulse-acquired sequence (bandwidth, 5 kHz; 2048 spectral points; excitation pulse, 10°; repetition time, 2 seconds; section thickness, 12 mm; 96 repetitions). ¹³C data collection was triggered by the Hypersense system (Oxford Instruments) to dynamically monitor the fate of hyperpolarized pyruvate after 200-μL injection via tail vein catheter. The catheter was made of 2-F silicone tubing, trimmed to reach from isocenter to the end of the magnet bore, with a total dead volume of approximately 100 μL. Injections were administered over approximately 11 seconds and were completed by approximately 21 seconds after dissolution.

Spectra were acquired by using a nominal excitation angle of 10° at intervals of 2 seconds. The half-height full-width area under spectral peaks for hyperpolarized [1-¹³C]-pyruvate and hyperpolarized [1-¹³C]-lactate were calculated to quantify their relative concentration at each time. Dynamic signal curves for hyperpolarized pyruvate and lactate were then integrated in time to calculate the total signal measured over the observable lifetime of each metabolite. The hyperpolarized lactate signal was normalized to the total ¹³C signal (hyperpolarized lactate and pyruvate) to correct for delivery and quantify the fraction of metabolically active hyperpolarized signal (precursor and product) observed in tumor that arises from the product lactate; this is referred to as the normalized lactate ratio [lactate/(lactate + pyruvate)] (J.L., a postdoctoral fellow with >2 years of experience, and J.A.B., associate professor in imaging physics with >15 years of experience). For anatomic tumor measurement, with software (Image J; National Institutes of Health, Bethesda, Md) the periphery of the tumor on T2-weighted MR images was manually traced on axial images to define a region of interest on each section (M.K.R., research scientist with >12 years of experience), and the area of the enclosed region was calculated. The area was then multiplied by the section thickness to obtain the volume of the object of interest within a section. To avoid overestimation of tumor size, half of the volume from the most dorsal and most ventral images that contained tumor was used in the volume calculations. Assuming a tumor density of 1 g/mL, tumor volumes (in cubic millimeters) were converted to weight (in grams) for analysis, as previously described (12).

PET/CT Imaging

For PET/CT imaging, anesthetized mice were injected intravenously with 150 μCi of ¹⁸F-FDG. Thirty minutes later, animals anesthetized with 2% isofluorane were imaged for 15 minutes by using a preclinical multimodal PET/CT scanner (Inveon; Siemens Healthcare, Erlangen, Germany). Images were reconstructed by using software provided by the manufacturer (Osem 3D; Siemens Healthcare). Regions of interest were placed in three planes (by M.K.R.) of each tumor to measure ¹⁸F-FDG uptake in percent injected dose per gram for each mouse by using software provided by the manufacturer.

Statistical Analysis

The primary comparison for the normalized lactate ratio was at 2 days. A linear mixed model (39 observations, three observations per mouse) was used to estimate and compare mean normalized lactate ratios between control and treatment group over time. Group and time nested within group were modeled as fixed effects and mouse as a random effect. The correlations between observations from the same mouse were accounted for by incorporating compound symmetric covariance structure into the modeling. Normalized lactate ratios were estimated and compared by using model estimates. All tests were two sided, and P values of .05 or less were considered to indicate statistical significance. Statistical analysis was performed by using statistical software (SAS version 9.4; SAS Institute, Cary, NC).

Results

Anatomic Evaluation of Efficacy

Longitudinal T2-weighted MR imaging revealed that treatment with pazopanib inhibited tumor growth in mice compared with the vehicle control (Fig 1, A), which was consistent with efficacy. Before therapy, there was no statistically significant difference in tumor weight between pazopanib-treated and vehicle control groups. Two days and 2 weeks after initiating treatment with pazopanib (n = 6), mice had significantly smaller tumors compared with control mice (n = 7; 0.054 g ± 0.041 [standard deviation] vs 0.223 g ± 0.112 and 0.142 g ± 0.063 vs 0.607 g ± 0.398, respectively; P < .05; Fig 1, B, Table E1 [online]). Tumor weight differences by necropsy confirmed these findings (0.205 g ± 0.058 vs 0.563 g ± 0.332; P < .05; Fig 1, B, Table E1 [online]). By comparing tumor size with calipers over time (Fig 1, C, Table E2 [online]), the size of tumors treated with pazopanib significantly increased at day 4 versus at day 2 (0.076 cm³ ± 0.040 vs 0.016 cm³ ± 0.005, respectively; P < .01; n = 6), then stabilized, and then increased in size on day 12 versus day 10 (0.255 cm³ ± 0.050 vs 0.067 cm³ ± 0.010 cm³, respectively; P < .001; n = 6), and at day 14 versus day 12 (0.322 cm³ ± 0.040 vs 0.255 cm³ ± 0.050, respectively; P < .05; n = 6). The data show that after day 10, pazopanib-treated tumors began to grow, which suggested less efficacy.

Figure 1:

In vivo MR imaging showed that pazopanib reduced the growth of established SKOV3 tumors. A, Representative axial T2-weighted ¹H anatomic MR images of mice before therapy (pre-therapy) or after 2 days and 2 weeks of therapy with vehicle (control mice) or pazopanib. B, Box plot shows tumor weight derived from serial MR images of the mice described in A before therapy or after 2 days and 2 weeks of therapy, or at necropsy (at 2 weeks after therapy; * P < .05). C, Graph of longitudinal caliper measurements. Pazopanib inhibited tumor growth compared with control mice (* P < .05). By comparing growth versus the day before in tumors treated with pazopanib, tumors grew on day 4 versus day 2, stabilized, and then grew on day 12 versus day 10 and day 14 versus day 12 (arrow represents P < .05). The error bars are standard deviation.

Hyperpolarized ¹³C MR Imaging

Intravenous administration of hyperpolarized ¹³C-pyruvate (200-μL injection, 80 mmol/L hyperpolarized ¹³C-pyruvate) resulted in readily detectable signal. As expected before therapy, no significant difference in ¹³C-pyruvate–to–¹³C-lactate conversion, observed by the normalized hyperpolarized lactate value, was observed in tumors between the prepazopanib therapy group and the control group (Figs 2, 3; Table E3 [online]). The flux of ¹³C-label from hyperpolarized ¹³C-pyruvate at 171 parts per million to ¹³C-lactate at 183 parts per million was increased in tumors 2 days after pazopanib therapy as suggested by the larger normalized area under the ¹³C-lactate curve (Fig 2, D). Two days after initiation of therapy, tumors from mice treated with pazopanib (n = 6) significantly increased conversion of ¹³C-pyruvate to ¹³C-lactate compared with before therapy (n = 7; 0.46 ± 0.07 vs 0.31 ± 0.14, respectively; P < .05). This was not observed in the control mice (Figs 2, 3). Complementing these findings, conversion of ¹³C-pyruvate to ¹³C-lactate was significantly increased 2 days after initiation of pazopanib therapy (n = 6) compared with control mice (n = 7; 0.46 ± 0.07 vs 0.35 ± 0.07, respectively; P < .05). No statistically significant difference was observed at 2 weeks.

Figure 2:

Graphs of dynamic MR imaging curves of tumor lactate and pyruvate peak intensities. Dynamic signal curves of MR spectroscopic imaging of hyperpolarized ¹³C-pyruvate and its metabolite ¹³C-lactate before (pre-therapy; A, B) or 2 days (C, D) and 2 weeks (E, F ) after control (A, C, E) or pazopanib (B, D, E) treatment. The mean signal for lactate and pyruvate was normalized to peak signal for each injection (inj ). Total hyperpolarized ¹³C signal was estimated by summing signal from hyperpolarized ¹³C-lactate and hyperpolarized ¹³C-pyruvate. s = seconds.

Figure 3:

Box plot shows ratio of lactate-to–pyruvate-and-lactate signal intensity. ¹³C-pyruvate–to–¹³C-lactate conversion before or 2 days and 2 weeks after control or pazopanib treatment. * P < .05.

¹⁸F-FDG PET/CT Imaging

No significant difference was observed regarding ¹⁸F-FDG uptake for the pazopanib-treated mice compared with control mice before therapy (n = 7; Fig 4, Table E4 [online]). There was not a significant difference after therapy between pazopanib-treated animals (n = 6) and control animals (n = 7) after 2 days or after 2 weeks.

Figure 4a:

¹⁸F-FDG PET/CT imaging did not indicate response to pazopanib treatment. (a) Representative coronal ¹⁸F-FDG PET/CT images of mice before or 2 days and 2 weeks after pazopanib or control treatment. B = bladder; K = kidney; T = tumor. (b) Box plot shows ¹⁸F-FDG uptake in percent injected dose per gram (%ID/g) before or 2 days and 2 weeks after pazopanib treatment or for control mice.

Figure 4b:

¹⁸F-FDG PET/CT imaging did not indicate response to pazopanib treatment. (a) Representative coronal ¹⁸F-FDG PET/CT images of mice before or 2 days and 2 weeks after pazopanib or control treatment. B = bladder; K = kidney; T = tumor. (b) Box plot shows ¹⁸F-FDG uptake in percent injected dose per gram (%ID/g) before or 2 days and 2 weeks after pazopanib treatment or for control mice.

Discussion

We found significant increase in hyperpolarized ¹³C-lactate signal relative to the ¹³C-pyruvate substrate signal in vivo in a SKOV3 human ovarian tumor model 2 days after starting therapy with pazopanib. This corresponded to tumor growth inhibition, which suggested that in the ovarian cancer model, hyperpolarized pyruvate imaging can serve as an early response indicator to pazopanib therapy. Data suggested greater enzymatic activity and/or amount of lactate dehydrogenase drives the process.

Interestingly, at 2 weeks, increased pyruvate-to-lactate conversion was not observed in pazopanib-treated tumors corresponding with regrowth of pazopanib tumors starting at day 10. Thus, both hyperpolarized pyruvate MR imaging and tumor growth data suggested decreased pazopanib efficacy at these later times.

In comparison, no difference was observed in tumors at ¹⁸F-FDG PET imaging between pazopanib-treated versus control-treated animals at either time. ¹⁸F-FDG PET imaging helps to show early events in glycolysis, whereas pyruvate-to-lactate conversion is a late event. Like glucose, ¹⁸F-FDG is brought into cells via glucose receptors and phosphorylated by hexokinase (primarily type 2), the first enzyme in glycolysis; in cancer, glucose transporters are upregulated and are often a more important factor than increased hexokinase activity (13) for greater ¹⁸F-FDG uptake. Phosphorylated ¹⁸F-FDG is not a good substrate for the next enzyme phosphoglucose isomerase; therefore, ¹⁸F-FDG imaging only represents early events of glycolysis. The altered metabolic phenotype of cancer, called the Warburg effect (14), includes increased activation of the glycolytic pathway even with aerobic conditions, including downstream increased lactate dehydrogenase activity and lactate production (15).

Because hyperpolarized MR imaging enables study of ¹³C-pyruvate and its metabolite ¹³C-lactate, lactate dehydrogenase activity can be assessed. Our data suggested that even when early glycolysis imaging is not informative, late events may still be informative regarding a response to a particular therapy. However, we do realize that as they are developed, more sensitive tests of early glycolytic events may show better correspondence with late events, including in our tumor model. Our data suggest that treatment with pazopanib preferentially affected late events in glycolysis measured by hyperpolarized pyruvate spectroscopy.

Regulation of metabolic activity is complex and can include altered cellular signaling and altered oxygen and nutrient supply. Pazopanib is a multityrosine kinase inhibitor that affects receptors found in tumor cells and cells involved in angiogenesis. Angiogenesis brings oxygen and nutrients to tumor cells and therefore regulates hypoxia and glycolysis. In gastrointestinal stromal tumors, the tyrosine kinase Kit drives growth and early glycolysis; therefore, these tumors have increased ¹⁸F-FDG uptake. Within 48 hours or 1 week of therapy with imatinib, which blocks Kit phosphorylation and turns off downstream signaling, ¹⁸F-FDG uptake was decreased in patients (16) and there was a corresponding decrease in expression of glucose transporters (17). However, gastrointestinal stromal tumors represent a special case where Kit activity is the primary tumor driver (18). Ovarian cancer is multifactorial and, as with most cancers, multiple pathways and receptors are involved. Pazopanib also inhibits angiogenesis.

Decreased pyruvate-to-lactate conversion has been suggested to be a marker of response in other models, for example colon cancer after radiation therapy (19), glioblastoma 7 days after mechanistic target of rapamycin inhibitor everolimus (20) or 1–2 days after alkylating agent temozolomide (21), and lymphoma 1 day after etoposide therapy (22). Decreased pyruvate-to-lactate conversion has also been noted at direct inhibition of glycolysis in p53-deficient lymphoma (23). These treatments directly affected tumor cells and resulted in decreased lactate production. However, we found increased ¹³C-pyruvate–to–¹³C-lactate conversion in our model. Tumor cell receptors affected by pazopanib, such as cKit (24) cFms (25), and fibroblast growth factor receptor (26,27) are not noted to be essential drivers of ovarian cancer cells. Platelet-derived growth factor receptor–α inhibition does not inhibit SKOV3 cell proliferation (28).

Pazopanib can affect vascular endothelial growth factor receptors on ovarian tumor cells and inhibit SKOV3 cell proliferation (29) but the effects in tumors are primarily paracrine and effect angiogenesis (30). Pazopanib inhibits ovarian tumor growth through its antiangiogenic activity and this may have led to an altered hypoxic state. This suggests that the antiangiogenic effect is the more likely cause of the increased lactate production in response to pazopanib.

Findings are supported by Fack et al (31) who injected ¹³C₆-glucose into mice with glioblastoma and performed metabolic flux analysis and immunohistochemistry in excised tissues; they found that the angiogenesis inhibitor bevacizumab (an antibody that binds vascular endothelial growth factor) increased glycolysis, increased lactate dehydrogenase activity, and lowered levels of Krebs cycle metabolites.

Our findings and those previously reported lead us to propose a model of differential responses observed by hyperpolarized ¹³C-pyruvate MR imaging depending on the mechanism of therapeutic response. In the proposed model, agents that effect tumor cell growth and metabolism signaling or result in DNA damage directly to tumor cells tend to decrease pyruvate-to-lactate conversion, whereas antiangiogenic agents affect the tumor microenvironment and alter hypoxia with resulting early increase in such conversion. This increase or decrease in pyruvate-to-lactate conversion may be evaluated in vivo by hyperpolarized ¹³C-pyruvate MR imaging.

Although the normalized lactate ratio is commonly thought to reflect lactate dehydrogenase activity, other factors may play a role, including perfusion (32), intracellular-reducing potential, and activity and number of monocarboxylate transporter 1, which is involved in bringing exogenous pyruvate into the cell (33). Hypoxia influences monocarboxylate transporter 1 (33) activity and lactate dehydrogenase activity (34). These are akin to glucose transporters and hexokinase at ¹⁸F-FDG imaging.

Differential uptake of ¹⁸F-FDG and ¹³C-pyruvate has been noted in a dog model, where the former was observed as uptake in muscle and liposarcoma, and lactate production observed preferentially in tumor (35). Note that ¹⁸F-FDG uptake is imaged commonly 30 to 90 minutes after injection, whereas ¹³C-pyruvate is imaged over approximately 2 minutes; there is a temporal disparity that can confound interpretation of differences between the two imaging techniques in addition to mechanistic differences of assessing early versus late metabolic events. However, increased glucose uptake may also be because of increased aerobic respiration, such as in muscle, brain, and sometimes inflammation. Clinically in ovarian cancer patients, Avril et al (36) noted that decreased uptake of ¹⁸F-FDG can suggest response to carboplatin-based therapy after one cycle in terms of progression-free but not overall survival. In a small phase I–II study, ¹⁸F-FDG PET was not found to be useful for detecting response in patients who received the angiogenesis inhibitor aflibercept (a decoy receptor that binds vascular endothelial growth factor) with docetaxel (37). Current findings suggest the measurement of ¹³C-pyruvate conversion to lactate may be more useful for evaluation of response to angiogenesis inhibitors, particularly pazopanib, in the setting of ovarian cancer. ¹³C-pyruvate MR imaging has been performed in clinical trials, which suggests potential for translation (38).

Limitations of our study include variability of measurements within tumors and between animals, which is common in animal models. The surface coil was centered on the tumor, but some normal tissue may have been included in the measurements. Also, cognates of clinical machines were used. Translation to humans may introduce technical problems, but a completed first-in-man study (38) provides some confidence.

In summary, our findings suggest that hyperpolarized ¹³C-pyruvate–to-lactate conversion can serve as an early indicator of response to multityrosine kinase inhibitor pazopanib in an ovarian cancer model and that evaluation of late events in glycolysis, such as by hyperpolarized ¹³C-pyruvate MR, may be predictive of response even when the imaging of early events, such as by ¹⁸F-FDG PET, is not predictive.

Advance in Knowledge

• ■ Findings suggest that hyperpolarized carbon 13 (¹³C)-pyruvate MR spectroscopy may serve as an early indicator of response to tyrosine kinase (ie, angiogenesis) inhibitors such as pazopanib in an ovarian cancer model (¹³C-pyruvate–to-lactate conversion, 0.46 ± 0.07 vs 0.035 ± 0.07; six mice with pazopanib, seven control mice; P < .05) even when fluorine 18 (¹⁸F) fluorodeoxyglucose (FDG) PET/CT imaging does not indicate a response.

Implications for Patient Care

• ■ Hyperpolarized ¹³C-pyruvate MR spectroscopy has the potential to serve as an indicator of tumor response even when ¹⁸F-FDG PET/CT does not indicate a response.