Multi-modality imaging to assess metabolic response to dichloroacetate treatment in tumor models


As increased glycolysis is a characteristic of various malignancies, reversing glycolytic metabolism is an appealing option for cancer therapy. Long used to treat lactic acidosis in a variety of disorders, dichloroacetate (DCA) has recently gained attention as a potential anti-cancer medication. By reactivating mitochondrial function and reducing glycolytic flow in tumor cells, DCA inhibits pyruvate dehydrogenase kinase, which causes cell cycle arrest and apoptosis. We recently showed that, while oxidative cells were less responsive to DCA treatment, DCA was able to selectively induce a metabolic transition in glycolytic cancer cells, resulting in a more oxidative phenotype and reducing proliferation. The purpose of the current investigation was to characterize the impact of DCA in glycolytic MDA-MB-231 tumors and in oxidative SiHa tumors utilizing advanced pharmacodynamic metabolic indicators in order to assess the applicability of this discovery in vivo. In tumor-bearing mice, oxygen consumption was quantified using 17O magnetic resonance spectroscopy, glucose uptake by 18F-FDG PET, and the conversion of pyruvate to lactate was assessed using hyperpolarized 13C-magnetic resonance spectroscopy. There was no discernible metabolic alteration in any tumor type. Surprisingly, all of these imaging metrics support the idea that both glycolytic and oxidative cancers responded to DCA similarly. These findings reveal a significant discrepancy in the bioenergetics of metabolic cancer cells between in vitro and in vivo settings, highlighting the crucial significance of the local microenvironment in tumor metabolic activities.

Keywords:tumor metabolism, hyperpolarized 13C-MRI, 17O MRS, and 18F-FDG PET


Many malignant tumors exhibit Warburg metabolism, which is accelerated glycolysis in the presence of oxygen and is connected to the aggressiveness, invasiveness, and poor prognosis of cancer. []. Targeting glucose metabolism is promoted as an alluring anticancer strategy with a high specificity and few unfavorable side effects because of the high glycolytic rate in different cancers. []. In fact, traditional therapies depend on the rapid proliferation mechanism that exists in both healthy and cancerous cells. Instead of targeting glycolytic metabolism, therapies should target metabolic alterations that support the Warburg malignant phenotype and are distinct from those found in healthy cells. Dedicated criteria for assessing tumor response to these novel therapies are urgently needed to facilitate medication development and evaluation in clinical trials. Moreover, using traditional anatomical imaging methods to measure therapy response for new cytostatic drugs that target tumor metabolism is not ideal. [] and only functional and molecular imaging methods might make it possible to determine the tumor response in advance. [].

Recently, we looked into how dichloroacetate (DCA) affected tumor cell lines with various metabolic profiles. []. Clinical trials have been successfully completed for the promising drug DCA, which encourages glucose oxidation over glycolysis by blocking mitochondrial pyruvate dehydrogenase kinase (PDK). []. We discovered that 5 mM DCA was more effective in cancer cells with a glycolytic phenotype, where a decrease in cell proliferation was caused by a reactivation of mitochondrial function and an increase in fluxes along the pentose phosphate and glycolytic pathways. Our findings suggested that patients with highly glycolytic malignancies might benefit from taking DCA. Determining the impact of DCA in these prototype tumor models in vivo, specifically the glycolytic MDA-MB-231 human breast cancer model and the oxidative SiHa human cervical cancer model, was the goal of the current investigation. We achieved this goal by employing a multi-modality molecular imaging strategy and a number of pharmacodynamic metabolic indicators. Prior to and following DCA treatment in tumor-bearing mice, oxygen consumption was assessed by 17O magnetic resonance spectroscopy (17O MRS), glucose uptake was assessed by 18F-fluorodeoxyglucose positron emission tomography (18F-FDG PET), and pyruvate conversion to lactate was measured by hyperpolarized 13C-magnetic resonance imaging (hyperpolarized 13C-MRI). Surprisingly, the previously observed in vitro behavior was not replicated in in vivo animals.


Oxygen consumption was measured to determine how DCA treatment affected the models’ in vivo metabolism (Figure (Figure1),1), glucose uptake (Figure (Figure2)2) and lactate flux (Figure (Figure3)3) were evaluated in MDA-MB-231 and SiHa tumors prior to DCA treatment and 24 hours later.


Figure 1

Effect of dichloroacetate on tumor oxygen consumption in vivo

Prior to, during, and following a 2-minute exposure to the 17O2 gas, representative MDA-MB-231 tumors (A) and SiHa tumors (B) were acquired for tumor H217O signals. The H217O signal is quantified as a percentage of the mean baseline signal preceding the administration of 17O2. Treatment with DCA has no effect on 17O2 metabolism. C. Evaluation of the H217O signal rate in tumors before and after 17O2 therapy. Means and SEM are used to express data. Tests in pairs have two sides..


Open in a separate window

Figure 2

Impact of dichloroacetate on in vivo tumor glucose absorption

Representative 18F-FDG PET scans depicting tumor-bearing mice models MDA-MB-231 A–B and SiHa C–D before and 24 hours after DCA treatment. Thin arrows designate tumors. The uptake of 18F-FDG is measured as %ID/g. The pictures were adjusted. The absorption of 18F-FDG in MDA-MB-231 and SiHa tumors is unaffected by DCA. E. Comparison of the uptake of 18F-FDG before and after therapy. Means and SEM are used to express data. Tests in pairs have two sides.


Open in a separate window

Figure 3

Effect of dichloroacetate on tumor lactate production in vivo

Representative MDA-MB-231 tumors (A–B) and SiHa tumors (C–D) hyperpolarized 1-13C pyruvate injection peak intensities of the tumor lactate and pyruvate. Before and after treatment, lactate production in MDA-MB-231 and SiHa tumors was assessed using the Lac/Pyr ratio E. Means and SEM are used to express data. Tests in pairs have two sides.

In Figure Figure1,1, For representative MDA-MB-231, tumor H217O spectra are shown (Figure). (Figure1A)1A) and SiHa tumors (Figure (Figure1B)1B) throughout 17O MRS tests. Under the same test settings, the results were quite repeatable. In MDA-MB-231 tumors, the evolution of the H217O signal, which illustrates the 17O2 metabolism in tumors, was the same before and after DCA treatment (Figure). (Figure1A)1A) and in SiHa tumors (Figure (Figure1B).1B). We discovered that DCA treatment had little to no effect on oxygen consumption in either tumor model, as determined by the rate of increase in the H217O signal (Figure). (Figure1C).1C). Prior to and following therapy, MDA-MB-231 tumors showed slopes of 1.02 10-3 0.16 10-3 s-1 and 0.91 10-3 0.09 10-3 s-1, respectively (n=5, P=0.5366). For SiHa tumors, the slopes were 0.79 10-3 0.11 10-3 s-1 after treatment (n = 5, P=0.2892) and 0.85 10-3 0.14 10-3 s-1 at baseline.

Figure shows the 18F-FDG uptake (%ID/g) measurements in both tumor models at baseline and after therapy. Figure2.2. We discovered that DCA treatment had no discernible impact on the uptake of 18F-FDG in either of the tumor models (Figure). 2A-2D), assessing a limited impact of DCA on glucose uptake (Figure (Figure2E).2E). For MDA-MB-231 tumors (n=7), the %ID/g measured on PET images (mean SEM) was 1.88 0.12 at baseline and 1.78 0.11 after treatment, and for SiHa tumors (n=7), the %ID/g measured on PET images was 1.79 0.21 at baseline and 1.89 0.19 after treatment.

Following hyperpolarized 1-13C pyruvate injection during hyperpolarized 13C-MRS experiments, the effect of DCA therapy on pyruvate transformation into lactate was assessed (Figure). (Figure3).3). Figure illustrates representative lactate and pyruvate peak intensities over time of MDA-MB-231 tumors and SiHa tumors that were photographed before and 24 hours after DCA treatment. 3A-3D. Lactate production was reduced after DCA treatment only in SiHa tumors (Figure (Figure3E).3E).In MDA-MB-231 tumors (n=7, P=0.3105), the lactate/pyruvate ratio (Lac/Pyr) changed from 0.55 0.05 to 0.48 0.04; in SiHa tumors (n=7, P=0.0348), it changed from 0.82 0.05 to 0.63 0.07.

By comparing the variation in the aforementioned biomarkers between the pre- and post-treatment situations, we were able to assess the treatment’s effectiveness. (Figure (Figure4).4). During treatment, there were no differences between MDA-MB-231 and SiHa tumors in terms of oxygen consumption or lactate flux measurements (P>0.05).(Figure 4A, 4C). For 18F-FDG uptake measurements, only a very slight but significant difference in behavior was seen (P=0.0240). (Figure (Figure4B).4B). Compared to SiHa tumors (n=7, +6.7 3.1%), 18F-FDG uptake was lower in MDA-MB-231 tumors following therapy (-5.4 3.5% vs. +6.7 3.1%, respectively).


Figure 4

Based on measurements of in vivo 17O2 metabolism, 18F-FDG uptake, and pyruvate conversion to lactate, DCA has little effect on the metabolism of glycolytic tumors compared to oxidative tumors.

Only a slight variation in behavior is seen for 18F-FDG uptake, and the magnitude of response to dichloroacetate (variation) is the same in both models. Means and SEM are used to express data. Unpaired tests have a two-sided outcome.


Using molecular imaging in vivo, the effect of DCA on tumors with various metabolic profiles was assessed in this work. Recent research has shown that DCA damages glycolytic cells more severely than oxidative cells. Using the same prototypical tumor models as in our in vitro study, the objective of the current study was to determine the applicability of these findings in vivo. [], specifically, the MDA-MB-231 human breast cancer model, which was found to be glycolytic. [] and the human cervical cancer SiHa model has been shown to be oxidative []. Earlier studies attesting to the effectiveness of DCA in tumors were used to determine the dose and administration method. [].

We previously found distinct effects of DCA treatment on glycolytic MDA-MB-231 human breast cancer cells’ oxygen consumption, glucose consumption, and lactate uptake in vitro. On the other hand, DCA treatment had no effect on the metabolic activity of oxidative SiHa human cervical cancer cells. We were unable to replicate these results in vivo using a multi-modality imaging experiment. MDA-MB-231 and SiHa tumors display the similar metabolic profile before therapy. As baseline hypoxic conditions were previously described for MDA-MB-231 and SiHa tumors [], Here, we focused on the fact that both tumor models display a glycolytic phenotype in anaerobic environments. Glycolytic MDA-MB-231 tumors post-treatment don’t seem to have been affected more than oxidative SiHa tumors. (Figure 1-4). Additionally, a few minor metabolic changes were discovered, including a significant reduction in lactate production in SiHa tumors. (Figure (Figure3E)3E) or a decreased 18F-FDG uptake in MDA-MB-231 tumors (Figure (Figure4B).4B). These findings taken together failed to show a definite metabolic change in MDA-MB-231 tumors or SiHa tumors treated with DCA for 24 hours. (Supplementary Figure S1). It was impossible to determine any treatment response in vivo due to differences in the growth rates of the different tumor models used in the study. (Supplementary Figure S2). Additionally, the repeatability of the measurements was previously established using the same tumor models.[]. Our research showed that the tumor metabolic response to DCA differed significantly in in vivo and in vitro settings.

DCA has been widely used to treat acquired or congenital forms of lactic acidosis because of its good tolerance and safety. []. When Bonnet and colleagues looked into DCA’s effects on cancer, they found that it promoted apoptosis in vitro and slowed tumor growth in vivo. []. Since then, this easily ingested and inexpensive molecule has undergone additional in vitro and in vivo research and successfully completed clinical trials. The earliest data from clinical trials show that DCA seems effective in adults with solid and brain tumors. []. However, in advanced non-small cell lung cancer, no conclusive findings stand out. []. In a different recent study by Feuerecker and colleagues, neuroblastoma tumors treated with DCA showed evidence of promoting tumor growth. [].These results suggest that the way different tumor types respond to DCA treatment can be very different.

The generic mechanism of action of DCA was initially postulated to involve the redirection of glucose metabolism from glycolysis to oxidation, which inhibits proliferation and induces caspase-mediated apoptosis. Reduced 18F-FDG uptake was seen after DCA therapy in a recent phase I study in patients with advanced solid tumors, supporting the use of 18F-FDG uptake as a potential biomarker of response to DCA.[]. Several preclinical studies have already used hyperpolarized 13C-pyruvate MRI to track the impact of DCA on solid tumors. [], but also in cardiac [] and brain studies []. In the present study, 18F-FDG uptake was unchanged after DCA treatment (Figure (Figure2E).2E). This shows that while a DCA treatment does not affect glucose absorption or phosphorylation, it may have an effect on how glucose is transformed afterwards. However, no modifications in the production of bicarbonate that would have indicated a switch in the energy metabolism from glycolysis to oxidative phosphorylation were seen. (Supplementary Figure S3). Recent research indicated that DCA might potentially function through additional methods. While it has already been proposed that a possible disruption of the equilibrium between fatty acid -oxidation and glucose oxidation may be another mechanism underlying the overall effects of DCA in vivo (as reviewed by []), PDK inhibitors might also trigger compensatory mechanisms that would lessen the effect of the medications on the overall metabolism of the tumor. Depending on the medication concentration and dosage regimen, DCA’s anti-cancer activities appear to use a variety of pathways. [] and cell type []. Additionally, a difference in PDK isoform expression between an in vitro and an in vivo model may have a significant impact on how DCA affects tumor metabolism in vivo. In fact, extracellular acidosis and the regulation of oncogenes in the tumor microenvironment can influence the expression of the PDK isoform. [], resulting perhaps in the expression of a PDK isoform less susceptible to the effects of DCA. Our results support a recent study that suggests the tumor microenvironment may influence the tumor phenotype just as significantly as the (epi) genetic profile. []. To ascertain the effects of DCA on energy metabolism in vivo, additional studies using pertinent isogenic cell clones with the capacity to form tumors in vivo should be taken into consideration.

Conclusion: In cancer models with different metabolic profiles, our multi-modality imaging study found significant discrepancies between in vitro and in vivo metabolic responses to DCA treatment. The findings indicate that implanted tumors and spontaneous cancer models should be preferred when studying DCA treatment in the context of the tumor microenvironment. Overall, more research is needed to determine how different tumor microenvironments affect the metabolic effects of DCA and their relevance for clinical use.



MDA-MB-231 (human breast cancer) and SiHa (human cervix squamous cell carcinoma) cell lines (American Type Culture Collection [ATCC]), were routinely cultured in Dulbecco’s modified Eagle’s medium containing 4.5g/l glucose supplemented with 10% fetal bovine serum and 1% penicillin-streptomycin.


American Type Culture Collection [ATCC] cell lines MDA-MB-231 (human breast cancer) and SiHa (human cervix squamous cell carcinoma) were routinely cultivated in Dulbecco’s modified Eagle’s medium containing 4.5g/l glucose supplemented with 10% fetal bovine serum and 1% penicillin-streptomycin.


Trypsinization was used to collect 107 MDA-MB-231 or 107 SiHa cells that had been amplified in vitro. After three rounds of washing with Hanks balanced salt solution, the cells were resuspended in 200 L of a 1:1 mixture of Matrigel (BD Biosciences) and Hanks balanced salt solution. Tumor cells were injected subcutaneously into the rear thigh of naked NMRI female mice for 17O MRS and PET scan investigations (Janvier Le Genest-Saint-Isle, France). The tumor cells were subcutaneously injected into the rear thigh of athymic nude female mice for hyperpolarized 13C-MRI investigations (Frederick Cancer Research Center, Animal Production, Frederick, MD, USA). When tumors were 7 mm in size, the studies were carried out because necrosis was less than 5% at that size, according to Hematoxylin and Eosin staining.

All animals underwent imaging prior to and following treatment, with a day in between each measurement, in order to evaluate the effects of DCA on tumor metabolism using biomarkers. The animal was given 200 mg/kg of dichloroacetate sodium (Sigma-Aldrich) intraperitoneally following baseline measurements. After the initial dose injection, another dose was administered. One hour following the administration of the drug, post-treatment measures were started. This delivery strategy is in line with earlier research employing hyperpolarized 13C-MRI to demonstrate the effects of DCA in tumors. []. The imaging protocol is summarized in Figure Figure55.


Figure 5

Experimental protocol

Mice were put to sleep by breathing in a continuous flow of air and isoflurane (2 L/min; Forene, Abbot, England). During the period of anesthesia, the animals were warmed (to about 35°C).


For the investigations on oxygen consumption, an 11.7 T (Bruker, Biospec) controlled by Paravision 6.0 was used for 17O MRS (Bruker, Ettlingen, Germany). A 1H/17O Bruker surface coil system was used for the experiments, which were performed across the tumor bulk. First, T2-weighted axial turbo RARE sequence (TR = 2500 ms; TE = 30 ms; rare factor = 8; NA = 2; FOV = 25 x 25 mm2; resolution: 98 x 98 m2; 1 mm slice thickness) was used to acquire the necessary anatomical images. A nonlocalized, single-pulse sequence (TR = 16.5 ms; NA = 600; repetition: 120; Tacq = 20 min; Acq BW = 5000 Hz; FA = 20°) was used to conduct the 17O MRS measurements. On samples of H217O with natural abundance, the 90° reference pulse for 17O MRS sequence was previously tuned.

A total of 120 17O-spectra were gathered in around 20 min, before, during, and after a 2 min inhalation time of the 17O2 combination, to calculate the tumor’s oxygen consumption during the 17O2 delivery. Using a custom Matlab software, we measured the integrals of the H217O peaks with time (The MathWorks Inc., Natick, MA, USA). After then, the H217O signal was expressed relative to the average baseline signal obtained before 17O2 administration. Between 1100 and 1200 seconds were used to calculate the mean signal of the final steady state (sfinal) during the post-inhalation interval. When the signal’s final value was between sfinal and 5% of signal variance, we believed the steady state had been attained. Between 600 seconds and the time when steady state was established, the slope during the linear incorporation phase was measured.


On a special small-animal PET scanner (Mosaic, Philips Medical Systems, Cleveland, USA) with a spatial resolution of 2.5 mm, whole-body PET imaging was carried out (FWHM). Whole-body acquisitions utilizing a helical CT scanner (NanoSPECT/CT Small Animal Imager, Bioscan Inc., DC, USA) were done after the PET scans. Anesthetized mice received a 120 l intraperitoneal injection of 11.1–14.8 MBq of 18F-FDG for each breathing condition (Betaplus Pharma, Brussels, Belgium). A 370 MBq 137Cs source was used to obtain a 10-minute transmission scan in a single mode for attenuation correction. After a subsequent 60-minute resting period, a 10-minute static PET acquisition was carried out. Images were reconstructed using a completely 3D iterative technique (3D-RAMLA) in a 128 x 128 x 120 matrix, with a voxel size of 1 mm3, following the correction with attenuation factors derived from the transmission scan. Anesthetized animals were moved from the PET scanner to the CT scanner (x-ray tube voltage: 55 kVp; number of projections: 180; exposure time: 1000 ms) on the same bed following PET acquisition in order to obtain anatomical references. The voxel size used to reconstruct the CT projections was 0.221 x 0.221 x 0.221 mm3. With the aid of PMOD software (PMODTM, version 3.403, PMOD technologies Ltd, Zurich, Switzerland), regions of interest (ROIs) were identified on PET scans. In order to create a 3D Volume of Interest (VOI) surrounding the tissue of interest, 2D ROIs were created on successive transversal slices using a 50% isocontour tool (ROI encompassing the pixel values larger than 50% of the maximum pixel). In order to distinguish hot pixels emanating from the surrounding tissues, such as the urine bladder, hot pixels from PET/CT fused images were used to prevent overestimation of the uptake within the VOI. The global tracer uptake in tumors was evaluated using the mean uptake within this VOI and reported as a percentage of the injected dose per gram of tissue (%ID/g).


According to the manufacturer’s instructions, 1-13C pyruvic acid (30 L), which contained 15 mM OXO63 and 2.5 mM gadolinium chelate ProHance (Bracco Diagnostics, Milan, Italy), was hyperpolarized at 3.35T and 1.4K using the Hypersense DNP polarizer (Oxford Instruments, Abingdon, UK). A superheated alkaline buffer containing 50 mM Tris(hydroxymethyl)aminomethane, 75 mM NaOH, and 100 mg/L ethylenediaminetetraacetic acid was used to quickly dissolve the hyperpolarized sample after 60–90 min. An intravenous injection of the hyperpolarized 1-13C pyruvate solution (96 mM) was administered via a catheter inserted into the mouse’s tail vein (12 L/g body weight). On a 3T scanner (MR Solutions, Guildford, UK), hyperpolarized 13C MRI experiments were carried out utilizing a home-made 13C solenoid leg coil. After the fast infusion of hyperpolarized 1-13C pyruvate, spectra were collected using a single pulse sequence every second for 240 seconds. The lactate/pyruvate ratio, which is derived from the areas under the curves of the 1-13C lactate peak and the 1-13C pyruvate peak, was used in a model-free approach to data analysis. [].


Utilizing the GraphPad Prism 7 program, analysis was carried out. Results are presented as the parameter’s mean value (SEM). Each and every statistical test has two sides. For each tumor model, the baseline and post-treatment mean changes were compared using a paired t-test, and the baseline and post-treatment mean changes were compared using an unpaired t-test. Results were deemed statistically significant if they had a P value of 0.05, 0.01 or 0.001 (*), respectively.

Leave a Reply