Cancer and Metastasis Reviews

, 27:545

Imaging of tumor glucose utilization with positron emission tomography


  • Andrea Buerkle
    • Department of Nuclear MedicineUniversity of Freiburg
    • Department of Nuclear MedicineUniversity of Freiburg

DOI: 10.1007/s10555-008-9151-x

Cite this article as:
Buerkle, A. & Weber, W.A. Cancer Metastasis Rev (2008) 27: 545. doi:10.1007/s10555-008-9151-x


In recent years, imaging of tumor glucose metabolism with positron emission tomography and fluorodeoxyglucose (FDG-PET) has become a routine test for detection, staging and restaging of malignant lymphomas and many solid tumors. FDG-PET is also increasingly used to monitor the effects of chemotherapy. The success of FDG-PET in oncologic imaging has generated considerable interest in understanding the molecular mechanisms underlying the markedly accelerated glucose use of almost all human cancers. Recent studies have indicated that there may be a close relation between the activation of oncogenic signaling pathways and cellular glucose utilization. For example deregulation of Akt, ras and MYC as well as loss of p53 function have been reported to confer increased glucose metabolic rates in cancer cells. These findings suggest that imaging of tumor glucose utilization may represent a marker for the activity of oncogenic pathways and metabolic changes during therapy may be used as a readout for the effectiveness of drugs targeting these pathways. However, the mechanisms for increased glucose metabolic activity of cancers cells are multifactorial and clinical studies will be necessary to determine in which context imaging of tumor glucose metabolism may be used for treatment monitoring.


Glucose metabolismCancerFluorodeoxyglucosePositron emission tomography

1 Introduction

Imaging of tumor glucose metabolism with positron emission tomography and the radiolabeled glucose analog [18F]2-fluorodeoxyglucose (FDG-PET) has been highly successful in recent years. Numerous studies have indicated that for most malignant lymphomas and many solid tumors FDG-PET provides a higher sensitivity and specificity for tumor staging than morphologic imaging modalities. Furthermore, FDG-PET is increasingly used for monitoring the effects of chemotherapy and now plays a central role in the assessment of tumor response in patients with malignant lymphomas [1]. Since 2000 the number of FDG-PET scans has increased about ninefold, and it is estimated that in 2007 more than 1.8 million studies have been performed in the US [2]. This remarkable success of FDG-PET is surprising, since virtually all cells of the human body consume glucose and it would therefore seem unlikely that imaging such a common process should allow sensitive and specific detection of tumor tissue. In fact, FDG was not developed for cancer detection and staging, but for studies of cerebral and myocardial glucose metabolism. Many years after the synthesis of FDG, nuclear medicine focused on imaging cancer with various antibodies binding to “tumor specific” antigens. Although there is a clear rationale for this approach, imaging with radiolabeled antibodies has so far only found only limited clinical applications. On the other hand, millions of FDG-PET studies have now demonstrated that the overwhelming majority of malignant tumors demonstrate markedly accelerated rates of glucose use. This unexpected observation has now stimulated considerable interest in understanding the molecular mechanisms underlying the altered glucose metabolism of cancer cells. Thus an old hypothesis by Otto Warburg is currently being revisited by many investigators. Warburg had proposed at the beginning of the twentieth century [3] that cancer was caused by altered cellular glucose metabolism.

Several recent studies have now indicated that activation of common oncogenes is associated with increased glycolytic rates. In this review we first provide a brief overview on the cellular uptake and metabolism of FDG and glucose and the technical aspects of imaging glucose metabolism with FDG-PET. We then summarize the findings of recent studies investigating the molecular mechanisms for the accelerated glucose metabolic rates of cancer cells. The review concludes with an overview of the consequences for the clinical use of FDG-PET. The individual applications of FDG-PET for tumor staging, re-staging and treatment monitoring have been extensively reviewed elsewhere [48] and are therefore not discussed in this review.

2 Transport and metabolism of glucose and FDG

Glucose metabolism is a key cellular function for maintenance of energy supply, and, more specifically, for supply with adenosine 5′-triphosphate (ATP), a universal energy currency that is essential for cell survival and proliferation. Under conditions of sufficient oxygen supply, glucose normally is metabolized in two sequential steps. First, glucose is metabolized to pyruvate in a process called glycolysis, which does not require oxygen. Glycolysis converts one molecule of glucose into two molecules of pyruvate, and generates energy in the form of two net molecules of ATP. Pyruvate then enters the mitochondria, where it is further metabolized in the citric acid cycle (Krebs cycle). Under low oxygen conditions, pyruvate is metabolized into lactic acid by lactate dehydrogenase (LDH) in a process called anaerobic glycolysis, which results in a relatively low yield of two ATP per glucose molecule. The yield for one glucose molecule being fully oxidized into carbon dioxide during aerobic metabolism is 19 times higher than for anaerobic metabolism, resulting in a total yield of approximately 30 ATP [9].

Cancer cells however do not follow these basic principles of glucose metabolism. Otto Warburg was the first to report [3] that tumor cells utilize more glucose and produce more lactic acid than normal cells. Even with sufficient oxygen supply tumor cells preferentially generate energy using glycolysis followed by metabolism of pyruvate into lactic acid and bypassing the mitochondrial oxidative phosphorylation. On the basis of these data he proposed that mitochondrial dysfunction is responsible for this uneconomical way for ATP production by cancer cells and hypothesized that a mitochondrial defect represents the key mechanism for tumor development. The “aerobic glycolysis” with increased glucose fermentation leading to lactate production and decreased oxidative phosphorylation in cancer cells therefore is also termed the “Warburg effect”.

Following intravenous injection FDG is transported across the cell membrane by sodium independent, facilitative glucose transporters (Gluts). These transporters allow energy independent transport of glucose across the cell membrane down a concentration gradient [9]. Thirteen members of the mammalian facilitative glucose transporter family have been identified. The genes belong to the solute carrier 2A family [9]. In malignant tumors Glut-1 is frequently overexpressed, but expression of Glut-3 and more recently Glut-12 has also been reported in several tumors types [9].

Unlike glucose FDG is not a substrate for the sodium dependent glucose transporters found in the tubulus system of the kidneys. As a consequence FDG is not reabsorbed after glomerular filtration, but excreted with the urine (Fig. 1). This contributes to the rapid clearance of FDG from the blood stream which is important for imaging metabolically active tissues with high contrast (Fig. 1).
Fig. 1

(a) Normal distribution of FDG. (b) Increased metabolic activity in multiple lesions in a patient with diffuse B-cell lymphoma. (c) Typical time course of FDG-uptake by malignant tumors and FDG clearance from the blood

Intracellularly, FDG and glucose are phosphorylated by hexokinase to glucose-6-phosphate and FDG-6 phosphate, respectively. Glucose-6 phosphate is then further metabolized to fructose-1,6-biphosphate and enters glycolysis. Alternatively, Glucose-6-phosphate enters the pentose phosphate pathway (hexose monophosphate pathway) and is converted to ribose-5-phosphate, which can serve as a building block for DNA and RNA synthesis. In contrast to this complex metabolic fate of glucose-6-phosphate, FDG-6-phosphate cannot be further metabolized in the glycolytic pathway because the F atom at the C2 position prevents FDG-6P from further degradation. Furthermore, tumor cells generally demonstrate a very low or absent activity of glucose-6-phosphatase, which catalyses the dephosphorylation of glucose-6-phosphate and FDG-6-phosphate. As the cell membrane is impermeable for phosphorylated FDG, FDG-6 phosphate becomes trapped and steadily accumulates in metabolically active tumor cells (Fig. 1).

While FDG only undergoes the first two steps of glucose metabolism (transport and phosphorylation by hexokinase), it is important to note that FDG uptake rates nevertheless reflect exogenous glucose utilization of cancer cells, as long as patients are imaged at steady-state conditions. At steady state conditions the concentrations of the various metabolites produced during glycolysis are constant and the net glucose flux across the cell membrane equals total exogenous glucose utilization. Consequently, uptake rates of FDG are not only dependent on the activity of glucose transporters and hexokinase, but also on the activity of downstream molecules, such as phosphofructokinase and pyruvate kinase, two important regulators of glycolytic activity [10, 11].

3 Quantification of tumor FDG uptake

The FDG concentration measured in a tissue by PET is the sum of three components: phosphorylated intracellular FDG, nonphosphorylated intracellular FDG, and nonphosphorylated intravascular FDG. Only the first component, the amount of phosphorylated FDG, is directly related to the metabolic activity of tumor cells. Static measurements of FDG uptake, which cannot differentiate these three components of the PET signal, therefore are not necessarily correlated with glucose metabolic rates. In order to measure metabolic rates of FDG it is necessary to image the time course of FDG uptake by the tissue and its clearance from the blood (Fig. 1). Using various tracer kinetic models it is then feasible to calculate the net FDG flux in the tumor tissue. However, this requires the acquisition of multiple images of the tumor region for about 1 h. Since the axial field of view of the PET scanner is about 15–20 cm these scans are only of limited use for tumor staging. Therefore, tracer kinetic analysis is rarely used to quantify tumor FDG uptake in clinical studies [8]. Instead FDG uptake is generally quantified by so called standardized uptake values (SUVs). The SUV represents the activity concentration in the tumor tissue at a fixed point in time, normalized to the injected FDG dose and the body weight of the patient. The basic concept underlying the use of the SUV is that at sufficiently late times after injection the concentration of intravascular FDG and nonphosphorylated intracellular FDG become so small that they can be neglected when compared to the concentration of phosphorylated FDG in the tumor tissue. In this case it can be shown that the net rate of FDG phosphorylation is directly proportional to the SUV [12]. Although SUVs can easily be obtained from routine whole-body PET scans, it is important to note that SUVs are influenced by a variety of factors other than tumor glucose use. These include among others the time between injection and imaging, the size of the lesion, image reconstruction parameters and the spatial resolution of the PET scanner [8]. Furthermore, FDG has a different affinity to glucose transporters and hexokinase than glucose. Therefore, metabolic rates for FDG are inherently different from glucose metabolic rates and it has been shown that the ratio between these two metabolic rates varies between different tissues. When considering all these factors it is evident that SUVs should not be considered as accurate measures of tumor glucose use. Nevertheless, SUVs have been found to be very useful for monitoring treatment induced changes in tumor glucose use over time. In this context, many of the confounding factors described above tend to cancel out. Studies have shown that the test–retest reproducibility of SUV measurements is high, if the PET studies are acquired according to a standardized protocol [8]. Relative changes in SUVs of about 20–25% can be reliably measured in individual patients and can be used as a criterion for a metabolic response to therapy [13].

4 FDG uptake, cellular proliferation and tumor grading

It has frequently been hypothesized that FDG uptake reflects tumor grading and cellular proliferation because less differentiated and more rapidly growing tumors are expected to metabolize more glucose for energy production and as a building block for biosyntheses. Although this hypothesis appears intuitive, the correlation between FDG uptake and cellular proliferation has been found to be weak in several studies. The most extensive studies have been performed in patients with non-small cell lung cancer. Vesselle et al. [14] and Buck et al. [15] found that tumor FDG uptake significantly correlates with the Ki-67 labeling index (p < 0.0001, N = 38 and p < 0.001, N = 26, respectively). While being statistically highly significant, the correlation between FDG uptake and Ki-67 labeling also showed a marked scatter. For example, in the study by Buck et al. [15] tumors with SUVs ranging from four to five demonstrated Ki-67 labeling indices ranging from 10% to 70%. Van Baardwijk et al. [16] and Yap et al. [17] found no significant correlation between tumor FDG uptake and the Ki-67 labeling index in a total of 124 patients. These discrepant findings may partly be due to differences in the range and distribution of the studied Ki-67 labeling indices as well as differences in the techniques for quantifying Ki-67 stained cells. However, comparative studies have shown a much closer correlation between uptake of the thymidine analog fluorothymidine and Ki-67 labeling [15]. Thus these technical factors alone cannot explain the lack of a close correlation between FDG uptake and cellular proliferation.

The correlation between tumor grading and FDG uptake has been extensively studied in multiple tumor types. In several studies a statistically significant correlation between FDG uptake and grading has been described, but generally the association was weak with considerable overlap between the individual tumor grades. Relatively strong correlations between tumor grading and FDG uptake have been reported in patients with gliomas and sarcomas [18, 19]. However, several benign or “borderline malignant” tumors, such as giant cell tumors [20], have been found to show intense FDG uptake. Similarly, juvenile pilocytic astrocytomas demonstrate high FDG uptake, despite their benign histopathologic appearance and favorable prognosis [21]. In summary current evidence suggests that cellular proliferation and differentiation influences FDG uptake, but it is also clear that the high metabolic rates of malignant tumors cannot be explained by increased proliferation or loss of cellular differentiation alone.

5 Expression of glucose transporters and hexokinase

Since Gluts and hexokinase mediate the uptake of FDG, multiple studies have correlated the expression levels of these proteins with FDG uptake. Several studies have indicated that Glut1 is overexpressed by cancer cells and that Glut1 protein expression levels as determined by immunohistochemistry correlate with FDG uptake [16, 2224]. However, some studies found no such association, but suggested that FDG uptake is correlated with the expression levels of hexokinase [25, 26]. Others found a statistically significant correlation with Glut1 and hexokinase levels [27], whereas yet others did find a correlation with Glut1, but not with hexokinase levels [28]. Furthermore, even in studies that found significant correlations between FDG uptake and hexokinase or Glut1 levels, correlations coefficients were frequently low. These unexpected and conflicting findings are probably to some extent explained by technical differences with regards to the acquisition and analysis of the PET scan as well as with regard to the techniques used for immunohistochemical staining and quantification of protein expression. Furthermore, there are three hexokinase isoforms and at least three sodium independent glucose transporters (Glut1, Glut3 and Glut12) that have been shown to be overexpressed in cancer [9]. Therefore, immunohistochemical staining for one transporter, such as Glut1 may be misleading. Perhaps even more importantly glucose transport and hexokinase activity are not only regulated on the protein level, but also by various other mechanisms. For example hexokinase activity has been shown to depend on the subcellular localization of the enzyme with mitochondrially bound hexokinase demonstrating markedly higher activity than cytosolic hexokinase [29]. Gluts transport glucose passively across the cell membrane. Therefore a rise in the intracellular glucose level will decrease glucose transport rates. These diverse regulatory mechanisms probably explain why there is no close correlation between Glut or hexokinase expression levels and FDG uptake.

6 FDG uptake and hypoxia

As a result of disordered and insufficient vasculature oxygen tension is frequently low in malignant tumors. This hypoxic environment selects for tumor cells that can adapt to oxygen deficiency. Gatenby and Gillies [30] recently proposed that early carcinogenesis occurs in a hypoxic microenvironment and that transformed cells initially have to rely on glycolysis for energy production. This metabolic adaptation also appears to also offer a proliferative advantage by suppressing apoptosis (see below). Furthermore, the ‘‘byproducts’’ of glycolysis (i.e., lactate and acidosis) contribute to the breakdown of the extracellular matrix, and increase the metastatic potential.

A key regulator of cellular response to hypoxia is hypoxia-inducible factor-1 (HIF-1). HIF-1 is a heterodimeric transcription factor, consisting of HIF-1α and HIF-1β subunits, which functions as a regulator of oxygen homeostasis, but also glycolysis [31]. Under regular oxygen tension, von Hippel–Lindau protein (pVHL), the product of the von Hippel–Lindau tumor suppressor gene (VHL), induces degradation of the transcription factor HIF-1. However, during hypoxia, or in cases of renal cell carcinoma that harbor mutations in the VHL gene, HIF-1 accumulates. Increased HIF-1 levels due to loss of function of pVHL lead to reduced mitochondrial mass by inhibiting mitochondrial biogenesis [32]. HIF-1 also induces pyruvate dehydrogenase kinase-1 (PDK-1) an enzyme which phosphorylates pyruvate dehydrogenase (PDH). This phosphorylation inhibits PDH activity and thus prevents pyruvate from being converted into acetyl-coenzyme A. Consequently, pyruvate generated by glycolysis cannot enter the Krebs cycle in the mitochondria and oxidative phosphorylation is reduced (Fig. 2) [33]. HIF-1 also transcriptionally activates genes encoding glucose transporters and glycolytic enzymes in response to reduced O2 availability [34]. Genes regulated by HIF-1 include glucose transporter Glut-1 and core enzymes of glycolysis such as aldolase A, phosphoglycerate kinase 1, pyruvate kinase M and lactate dehydrogenase A [3436]. Overall, HIF-1 functions as a metabolic switch that shunts glucose metabolites from the mitochondria to glycolysis to maintain ATP production under hypoxic conditions. In addition, HIF-1 is activated by various other signaling pathways and may therefore modulate glucose metabolism in response to activation of other oncogenes as discussed in the next section.
Fig. 2

Overview of the metabolism of glucose and FDG and some of the regulatory pathways for glucose metabolism in cancer cells

In cell culture, short term hypoxia (≤4 h) has been demonstrated to significantly increase FDG uptake of ovarian, melanoma and breast cancer cell lines. In these experiments increased FDG uptake was likely mediated by translocation of glucose transporters to the plasma membrane or changes in glucose transporter activity [37, 38]. In a recent clinical study HIF-1 expression as assessed by immunohistochemistry has been shown to correlate with FDG uptake in patients with non-small cell lung cancer [16]. However, no such association was found in a group of patients with breast cancer [22]. In several studies tumor FDG uptake has been correlated with the uptake of [18F]fluoromisonidazole (FMISO), a PET imaging probe, which is selectively retained in hypoxic cells. Overall, the correlation between FMISO and FDG uptake was found to be relatively weak [3941]. In a study including 26 patients with head and neck cancer Rajendran et al. [40] correlated the hypoxic tumor volume with maximum FDG uptake. The authors observed a significant correlation (p = 0.038), but the correlation coefficient was low (Spearman’s rho = 0.41). In the same study the authors found no significant correlation between the hypoxic volume and FDG uptake for patients with breast cancer and brain tumors [40]. The same group correlated FDG and FMISO uptake pixel-by-pixel in a group of 19 patients with soft tissue sarcoma. The mean correlation coefficient was 0.49 (range 0.09–0.79) indicating that regional hypoxia and FDG uptake frequently do not correlate.

7 Oncogenic signaling and accelerated glucose use

The previous sections have discussed how glucose metabolism of cancer cells is up-regulated secondary to cellular proliferation and in response to hypoxia. However, there is increasing evidence that important oncogenes also play a role in regulating cellular glucose metabolism. Their uncontrolled activity in cancer may therefore explain the glycolytic phenotype of most cancer cells. Various mechanisms for activation of glycolysis by oncogenes have been described. Some oncogenes increase the transcription of glycolytic genes. For example, the transcription factor c-Myc directly binds numerous glycolytic genes including those encoding hexokinase 2, enolase and lactate dehydrogenase A [42, 43]. Overexpression of glycolytic enzymes caused by unregulated c-Myc activity may therefore represent one mechanism for the increased glucose utilization of cancer cells.

Several oncogenic signaling pathways result in activation of HIF-1, thereby inducing a hypoxia response and resulting in increased expression of glucose transporters and glycolytic enzymes as described above. Growth factor signals shown to interact with HIF-1 include the epidermal growth factor (EGF), insulin-like growth factor and human epidermal growth factor receptor 2 pathways. In addition, activation of ras and src oncogenes can activate Hif-1 [4452]. Thus HIF-1 activation by either hypoxia or oncogenic signaling may represent a central mechanism for the increased glycolytic activity of cancer cells.

Oncogenic ras has been found to increase the concentration of fructose-2,6-bisphosphate (F2,6BP) [11, 53] an allosteric activator of the enzyme 6-phosphofructo-1-kinase (PFK-1) [10]. PFK-1 catalyzes the phosphorylation of fructose-6-phosphate (F6P) to fructose-1,6-biphosphate. This phosphorylation is the first committed step of glycolysis and represents an important regulator glycolytic activity. F2,6BP relieves the inhibition of PFK-1 by ATP and allows glycolytic flux at the PFK-1 checkpoint to proceed. F2,6BP levels depend on the activity of the enzyme 6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase (PFK2/FBPase), which is capable of phosphorylating F6P to F2,6BP and dephosphorylating F2,6BP to F6P. There are four isoforms of PFK2/FBPase (PFKB1-PFKB4), and the PFKFB3 isoform has been found to be overexpressed in lung, breast, prostate and colon adenocarcinomas [54]. Lung fibroblasts derived from PFKFB3+/− mice and transformed with SV40 T antigen and H-rasV12 exhibit decreased F2,6BP, glycolytic flux to lactate resulting in increased deoxyglucose uptake [11] as compared to transformed PFKB3+/+ mice. In addition to elucidating the mechanisms of ras transformation on cellular metabolism, this experiment also shows how enzymes downstream of hexokinase can affect deoxyglucose uptake. Treatment of glioblastoma cells with the ras inhibitor trans-farnesylthiosalicylic acid caused a downregulation of eight glycolytic enzymes resulting in reduced production of lactate and ATP [53].

The Akt family of protein kinase plays a major in multiple human diseases including diabetes and cancer. Akt is an integral part of the insulin receptor signaling and has a well established role in increasing glucose uptake and metabolism of insulin sensitive tissues [55]. However, Akt is also activated downstream of multiple growth factor receptors including receptor tyrosine kinases frequently involved in carcinogenesis, such as the EGF receptor kinase [56, 57]. Activated Akt has been shown to regulate a number of key apoptotic proteins including Bad. As a result the proapoptotic functions of Bad are suppressed and the expression of antiapoptotic proteins is increased [56]. Akt activation is therefore considered to provide an important survival signal for cancer cells. Recent studies have suggested that Akt is also involved in regulating glucose metabolism of cancer cells and that the antiapoptotic and metabolic functions of Akt are closely linked. By negatively regulating the activity of glycogen synthase kinase 3beta Akt induces the translocation of hexokinase to the mitochondrial membrane where it binds to the voltage-dependent anion channel, suppressing apoptosis [5759]. On the other hand mitochondrial binding is known to increase the enzymatic activity of hexokinase [29]. Thus, by promoting hexokinase binding to mitochondria Akt can provide an antiapoptotic signal and stimulate glycolysis. Hematopoietic cells transfected with a constitutively active form of Akt demonstrated increased glycolytic activity in the absence of increased proliferation [60]. Akt can also increase the expression of glycolytic genes by activation of HIF-1 [61] as demonstrated in a transgenic prostate cancer model in which Akt activity resulted in upregulation of HIF-1 targets including glucose transporters and glycolytic enzymes [61].

A recent study suggested that the tumor suppressor p53 also plays a critical role in glucose metabolism, stimulating mitochondrial respiration independently from its effect on the cell cycle [63]. Tumor cells with p53 loss display decreased oxygen consumption and increased lactate production. P53 can directly transactivate the gene for “synthesis cytochrome c oxidase 2” (SCO2) which is required for the assembly of the cytochrome c oxidase complex, an integral part of the respiratory chain. Consequently, loss of p53 or SCO2 results in a switch from cellular respiration to aerobic glycolysis [63]. Moreover, p53 also can negatively regulate the glycolytic enzyme phosphoglycerate [64] and Akt [65]. The importance of p53 is furthermore supported by the observations that loss of p53 increases FDG uptake of breast cancer cells [66], and, in clinical studies, immunohistochemical staining for p53 correlated with FDG uptake in patients with colorectal cancer [67].

The data summarized above describe several potential molecular mechanisms underlying the aerobic glycolysis of cancer cells. According to these studies altered glycolysis is an immediate consequence, but not a cause of malignant transformation. For some tumors, however, recent studies seem to support Warburg’s original hypothesis that metabolic alterations are in fact the cause of tumorigenesis. Overexpression of the glycolytic enzyme phosphoglycerate mutase (PGM) has been shown to immortalize mouse embryonic fibroblasts. PGM overexpression enhanced glycolytic flux, allowed indefinite proliferation, and rendered cells resistant to ras-induced arrest. Glucosephosphate isomerase, another glycolytic enzyme, displayed similar activity and, conversely, depletion of PGM or glucosephosphate isomerase with short interfering RNA triggered premature senescence [63].

Two mitochondrial enzymes, succinate dehydrogenase (SDH) and fumarate hydratase (FH) have been found to act as tumor suppressors and to be involved in the development of various benign and malignant lesions [67]. Both enzymes catalyze reactions in the citric acid cycle. SDH is a complex of four different polypeptides (SDHA–SDHD) and converts succinate to fumarate. FH catalyzes the next step of the citric acid cycle, the conversion of fumarate to malate [67]. Inherited or somatic mutations in SDHB, SDHC and SDHD have been shown to lead to the development of pheochromocytoma, whereas inherited mutations in FH can cause leiomyoma, leiomyosarcoma or renal-cell carcinoma [6770]. The mechanisms for the tumorigenic effect of mutations in SDH and FH have not been fully elucidated. One hypothesis is that mutations increase the productions of reactive oxygen species (ROS) in the mitochondria. ROS may cause oxidative damage to nuclear DNA leading to increased mutagenesis and eventual malignant transformation. In addition ROS can stabilize HIF-1α causing a “pseudo-hypoxic” state. HIF-1 activity may support tumor development by facilitating neovascularization and inhibiting apoptosis [67]. HIF-1α stabilization may also be caused by the accumulation of succinate in cells with SDH mutations. Degradation of HIF-1α requires hydroxylation catalyzed by the enzyme prolyl-hydroxylase (PHD). In this reaction alpha-keto-glutarate is converted to succinate. Accumulation of succinate reduces activity of PHD by product inhibition [71].

SDH and FH mutations have so far only been observed in relatively rare tumors, such as pheochromocytomas and a small subgroup of renal cancers. A recent paper has, however, suggested that mitochondrial dysfunction may be a common phenomenon in cancer [72]. In a variety of cancer cell lines mitochondria have been found to be hyperpolarized, i.e. the proton gradient across the mitochondrial membrane is higher than in normal cells [72]. This hyperpolarization is considered to induce a state of apoptotic resistance by inhibiting the efflux of proapoptotic mitochondrial factors. This hyperpolarization can be overcome by treating cancer cells with the PDK-1 inhibitor dichloroacetate (DCA). DCA treatment increases oxidative phosphorylation in cancer cells, reduces the mitochondrial membrane potential and induces apoptosis in cell culture and in xenograft models [72]. Oxidative phosphorylation of cancer cells can also be increased by knockdown of LDH-A [73]. In addition, LDH-A knockdown caused a decrease of the mitochondrial membrane potential, resulted in decreased tumorigenicity and compromised the ability of cancer cells to proliferate under hypoxia [73].

8 Consequences for glucose metabolism imaging in the clinic

As discussed above multiple mechanisms have been identified that lead to increased glucose metabolic activity of cancer cells. This suggests that attempts to attribute the increased FDG uptake of malignant tumors to the expression of one particular protein, such as Glut-1, cannot be successful, since FDG uptake reflects the integrated functions of glucose transporters and enzymes of the glycolytic pathway. Current evidence indicates that glycolytic flux is regulated at the level of glucose transport as well as by hexokinase and phosphofructokinase activity. At steady state conditions all these transport processes and enzymatic reactions affect FDG uptake.

Conversely, there seems to be no single cause for the increased FDG uptake of cancer cells. FDG uptake is modulated by the tumor environment (e.g. hypoxia), but also related to proliferation, activation of oncogenes and loss of tumor suppressor genes. This likely explains why FDG PET has been so successful for tumor staging. If, for example, FDG uptake only reflected hypoxia, it would be unlikely for FDG-PET to be a sensitive test for tumor staging, since many tumors are not particularly hypoxic. Likewise, FDG-PET could not be a sensitive test, if FDG uptake was caused by the activation of one specific oncogene. While the multifactorial causes of increased FDG uptake of cancer cells are therefore advantageous for tumor staging, they also prevent assessment of parameters other than glucose metabolism by FDG PET. Statistically significant correlations between FDG uptake and various other biologic characteristics of the tumor tissue have been observed in larger groups of patients, but the strength of the association appears far too low to assess for example hypoxia of an individual tumor on the basis of its FDG uptake. Similarly, it seems unlikely that the activity of oncogenic signaling pathways can be determined by measuring tumor FDG uptake. In some cases clinical data on tumor FDG-uptake are in fact difficult to reconcile with experimental data on oncogene activation. For example, aberrant HIF-1 activity due to loss of the tumor suppressor VHL is very well documented in renal cell carcinoma. Based on the extensive data on regulation of tumor glucose utilization by Hif-1 one would therefore expect renal cell carcinomas to be highly hypermetabolic. In animal models, VHL knockdown has been observed to increase FDG uptake of renal carcinomas [74]. However, clinically renal cell carcinomas are characterized by relatively low FDG uptake [75].

Nevertheless, changes in FDG uptake may potentially provide an early readout for the successful inhibition of a particular pathway by a targeted drug. In this case, other factors affecting tumor FDG uptake remain constant and a decrease in FDG uptake may indicate successful target inhibition. In patients with gastrointestinal stromal tumors, FDG uptake decreases within hours after treatment with the c-kit inhibitor imatinib [76, 77]. Experimental studies have suggested that this rapid reduction of glucose metabolic activity does not reflect cell death, but a translocation of glucose transporters from the plasma membrane to the cytosol [78]. Similarly, inhibition of EGFR kinase with gefitinib has been shown to result within 4 h in glucose transporter translocation in lung cancer cell lines with activating mutations of the EGF receptor (EGFR) kinase domain [79]. Future clinical studies will be necessary in order to determine whether similar rapid changes of glucose metabolism do also occur in patients treated with EGFR kinase inhibitors or other targeted drugs.

Finally, tumor glucose metabolism is increasingly considered to represent a promising therapeutic target, since cancer cells may be uniquely sensitive to inhibition of glycolysis [60, 73] or activation of mitochondrial respiration [72]. A challenge for this new therapeutic approach is the glucose dependency of the brain, which accumulates FDG to a similar extent as malignant tumors (Fig. 1). However, several approaches are currently being tested in animal experiments and phase I clinical trials [80, 81]. FDG-PET imaging provides the unique opportunity to select tumors for treatment with glycolysis inhibitors and to monitor their effectiveness.

9 Conclusions

Increased glucose use of cancer cells was first described at the beginning of the twentieth century, but the causes and mechanisms underlying this phenomenon are only starting to be revealed. Recent experimental data suggests that increased glycolytic rates of cancer cells are not only secondary to cellular proliferation and intratumoral hypoxia, but are also caused by genetic alterations in various oncogenic signaling pathways that result in resistance to apoptosis, shutdown of oxidative phosphorylation and activation of glycolysis. If these observations are confirmed clinically, FDG-PET may become a valuable tool for monitoring inhibition of these signaling pathways by targeted drugs.

Copyright information

© Springer Science+Business Media, LLC 2008