Molecular Imaging and Biology

, Volume 8, Issue 3, pp 141–150

Positron Emission Tomography Measurement of Tumor Metabolism and Growth: Its Expanding Role in Oncology


    • Karmanos Cancer Institute, Department of MedicineWayne State University
Review Article

DOI: 10.1007/s11307-006-0039-2

Cite this article as:
Shields, A.F. Mol Imaging Biol (2006) 8: 141. doi:10.1007/s11307-006-0039-2


This work highlights the explosion and evolution of positron emission tomography (PET) for use in oncology research and clinical practice. 2-Deoxy-2-[F-18]fluoro-d-glucose (FDG)-PET is important in the staging of cancer, estimation of prognosis, and for its ability to predict therapeutic outcome. A number of new imaging agents are under development and may find a place in oncology when studies prove their utility. This scientific overview includes a review of the development of a number of thymidine analogs, such as 18F-3′-deoxy-3′-fluorothymidine (FLT) and 18F-1-(2′-deoxy-2′-fluoro-beta-d-arabinofuranosyl)-thymine (FMAU), including chemical structure variations; their application in a variety of tumors; and the role of various kinetic models for understanding cellular proliferation. The greatest unmet need for PET is in further developing and validating its use in the measurement of treatment response.

Key words



Positron emission tomography (PET) is a relatively recent addition to the world of oncologic imaging compared to more standard X-ray radiography and even cross-sectional imaging obtained with computed tomography (CT) and magnetic resonance imaging (MRI). Although the last few years have seen an explosion in the use of PET for oncology research, this mature technology is still relatively immature for sophisticated clinical use. At this point, it is worthy to take stock of the present state of oncologic therapy, the present role of PET in oncology, and the future direction of both fields.

Oncologist's Use of PET Imaging

Physicians presently practicing in the field of oncology have learned, through their training and experience, to rely on imaging as part of their day-to-day care for cancer patients. In collaboration with radiologists and nuclear physicians, they use imaging to determine the extent of the tumor, making use of plain radiography, including chest X-rays, cross-sectional imaging with CT and MRI, and some bone scans. With these techniques, physicians measure the location and size of tumors, and how each of these changes with time. While physiologic activity within the tumor is required for the uptake of contrast in CT and MRI, or a metabolic agent utilized in bone scanning, measurement of tumor metabolism has not been a regular interest of oncologists. The availability of PET is changing this situation.

Oncologists have spent years looking at images obtained with standard anatomic imaging techniques. As a result of their special interest in particular tumors and the images obtained as part of staging or treatment evaluation, oncologists feel comfortable reading many images in collaboration with their colleagues in radiology. They are familiar with the strengths and limitations of these techniques. The sensitivities and specificities of anatomic imaging are well understood, and oncologists have experience in many of the common pitfalls encountered in imaging. Cross-sectional imaging provides a detailed look inside the body and it is rare to find a patient with a completely “normal” study. Few patients have flawless anatomic images just as no patient presents with a “normal” skin exam because everyone has moles, blisters, scars, and areas of altered pigment. Small irregularities that may represent cysts, scars, and anatomic variation are regularly seen. For example, in the ongoing CT study for lung cancer screening, about 70% of the patients have been found to have lesions within their lungs, while a small fraction subsequently are documented to have cancer [1]. Oncologists are also adept at recognizing many of the limitations found in the written reports that accompany the images. When small abnormalities are reported, the oncologist must determine the likelihood of a true pathologic finding, its effect on the treatment of the patient, the need for further follow-up of such lesions, and when the report can be safely ignored. Given the litigious climate in the United States, reports often contain descriptions of lesions that the oncologist feels are unlikely to be of significance and must be ignored.

When surgery was the only real effective modality in cancer care, presurgical staging was the main role of imaging. The extent of the tumor determined the surgeons' approach and whether surgery was appropriate at all. The myriad of options now available for the treatment of advanced, unresectable cancer have led to a growing importance in the measurement of a tumor's response to therapy. Anatomic measurements of changing tumor size have been the main component of such assessments using WHO and RECIST criteria. [2, 3]. PET is just beginning to gain a role in such measurement and this will be discussed further in this paper.

In the past few years, 2-deoxy-2-[F-18]fluoro-d-glucose (FDG)-PET has been primarily approved for use in the staging and restaging of several tumors. It is routinely used in the staging of lung cancer at many centers, and has found regular use in the assessment of colon, breast, head and neck, esophageal, melanoma, and lymphoma [4]. Despite the large number of published studies and metaanalyses addressing the use of FDG-PET in these tumors and others, many oncologists are just beginning to use PET. It is only in the last couple of years that FDG-PET has become available at many smaller medical centers and away from urban areas, and penetration varies greatly throughout the United States and across the globe. Even for well-documented and appropriate uses, e.g., staging of lung and esophageal cancer, the referring physician may not utilize it because of lack of acceptance. For some tumor types, where reasonable clinical data exists, e.g., pancreas cancer, PET may not be ordered because of cost or lack of reimbursement.

Importance of Treatment Assessment

Generally, oncologists wish to know when a treatment has failed, as identification will allow the oncologist to move onto the next therapy. Imaging provides information in evaluating treatment response, and accurate and rapid assessment is gaining more importance. Clinical oncologists need help in deciding when to abandon their primary therapeutic approach, now that new second- and third-line treatments have become available for those with advanced disease. It was not that many years ago that oncologists debated the use of any chemotherapy for metastatic, nonsmall cell lung cancer, as it has only recently become clear that such treatment leads to improvement in survival rate. In the last couple of years, second- and third-line treatments have come into routine use. Early and accurate measurements of response allow clinicians to avoid toxicity and high costs. While PET is one of the more costly imaging procedures, the costs of cancer treatment have been rapidly escalating. For example, treatment of a gastrointestinal stromal tumor with oral imatinib costs about $2000/month for an effective but palliative therapy. Even more extreme is the cost of cetuximab, which costs about $10,000/month, despite its response rate of only about 10% in patients with refractory colorectal cancer. Such costs make PET imaging very reasonable if it can eliminate ineffective therapy, and standard anatomic imaging can be greatly delayed in assessing treatment response. For example, some cancers grow very slowly, and it can take a number of months before a tumor has sufficiently increased in size to meet standard RECIST criteria for progression. On the other hand, in patients who ultimately have tumor regression, e.g., a 30% decrease in tumor diameter for RECIST criteria, it may take many months for the tumor cells to die, the dead cells to be cleared, and the inflammatory response to resolve. PET may provide added or faster information about treatment response.

The development of new therapeutic agents may also benefit from the use of new imaging approaches to drug assessment. New drug development is an exceedingly long and expensive process. Because most agents fail as they go through the process of development, many drug companies have the philosophy that they want to “fail fast.” This allows them to concentrate on more promising agents. When the overall study and drug development costs are added up, imaging may not add significantly to the costs. For example, in a study of the oncolytic adenoviral vector (ONYX-015, CI-1040), a phase II study was designed to include routine FDG-PET and CT imaging every two cycles [5]. In planning this trial, one option under consideration was to just perform PET imaging for patients with partial response or prolonged stable disease on CT, thereby allowing PET to help verify the anatomic imaging results. The company developing this new therapeutic agent wanted to gain any possible information that could indicate that this agent was effective, and chose to obtain more frequent PET scans, despite the added costs. It was noted that the development of this unique agent for phase III trials would require scale up of the manufacturing of this viral agent, and this would require many millions of dollars to build a new manufacturing facility. The added costs of FDG-PET were considered very minor compared to these development costs. When both PET and CT showed no evidence of tumor response in a phase II study of 18 patients, further study of this treatment was discontinued.

PET and the Staging of Cancer

Proper staging of patients is critical in determining the patient's prognosis and appropriate therapy, and imaging plays a role that complements data from pathology in determining a tumor's spread. This was first seen in studies on patients with lung cancer, where FDG-PET was found to be very useful in the staging of mediastinal nodes. In a metaanalysis, PET had a sensitivity and specificity of 84% and 89%, compared to CT scanning with a 57% sensitivity and 82% specificity [6]. Such studies have also found that PET will find evidence of stage IV disease in about 10% of the patients. The detection of stage IV disease will generally lead to palliative treatment in most patients with solid tumors, whereas stage III tumors often receive a combination of treatments, including surgery done with curative intent. Improved staging can also alter results in unexpected ways, as more accurate staging can result in the phenomenon known as stage migration or the “Will Rogers' phenomenon.” This is named after the homespun comedian from Oklahoma, who reportedly remarked that “when the Oakies left Oklahoma for California, they raised the average IQ of both states.” This effect is often seen when improved imaging detects small volumes of tumor in patients originally classified as stage III. Removal of patients with now detectable metastatic disease from the stage III group leads to improved survival in the stage. Similarly, the new patients who are grouped with other stage IV patients can result in overall improvement in this group, because one is now including patients with small volumes of metastatic tumor. While each new staging group may have improved survival, even without any change in therapy, the overall survival in the combined groups may remain unchanged. This is one reason for the requirement of randomized control trials, because improved imaging may make previous studies unusable for comparison. This effect was recently noted in patients with colorectal cancer and who had FDG-PET evaluation prior to resection of a limited number of liver metastases [7] (Fig. 1). A number of previous studies have shown that resection of patients with a few liver metastases can result in five-year survivals of about 30%. When PET staging is performed prior to resection, patients with more widespread, but undetected metastases are regularly removed from consideration of surgery. The group that was retained for surgery had an improved five-year survival of 58% (45–72%, 95% confidence interval). Similar effects may be seen with any test that alters the grouping of patients in a particular treatment category, including earlier detection with screening or new molecular techniques.
Fig. 1

Survival in patients with liver resections for metastatic colorectal cancer. Surgery was performed with curative intent. The graph includes survival curves from 13 published studies not using FDG-PET staging in comparison to overall survival in this series where PET was used to select resection candidates. From Fernandez et al. [7] (used with permission).

Prognostic Use of Metabolic Imaging

Obtaining accurate prognostic information is critical in determining the optimal treatment approach to individual patients. For example, data suggests that elderly patients with very early evidence of prostate cancer have a low risk of disease progression and spread in their lifetime [8], and may be observed without surgery or radiation. Patients with localized cancers of the lung, breast, and colon, and depending on prognosis as determined by tumor size, nodal status, and molecular markers, are regularly offered surgery combined with adjuvant chemotherapy. On the other hand, patients with particularly poor prognosis (e.g., those who have stage IV disease), can be spared aggressive surgery because it will not improve their outcomes, and palliative chemotherapy can be offered. FDG clearly adds to this information, as indicated above, by improving the accuracy of staging. The metabolic information provided by the scan may also include prognostic information. The author must admit that he was initially skeptical that the limited information from a single metabolic pathway, as imaging with FDG-PET, could augment estimates of disease prognosis that can be readily obtained by using extensive clinical and laboratory information. From the tumor pathology, one already obtains information regarding grade, size, extent of local invasion, and nodal involvement. Staging information is also obtained by routine anatomic imaging, carried out with CT and MRI and the recent addition of FDG-PET. General patient status, age, gender, weight loss, and performance status are also important prognostic indicators. Blood tests are obtained to measure organ function, including liver, renal, and marrow status, and serum markers, such as PSA for prostate tumors and alpha fetoprotein for hepatocellular cancers, are regularly used by clinicians. More recently, researchers have also begun to measure a number of molecular functions by using immunohistochemical staining of the tumor, and to study gene array and proteonomic analysis to assay thousands of tumor genes. PET imaging of tumor metabolism using a number of new radiotracers might be expected to not add significantly to estimates of prognosis given the growing range of tools already available. Studies continue to show that in vivo metabolism can be very worthwhile. A number of studies have shown that the simple measurement of standardized uptake value (SUV) provides a measure of tumor aggressiveness [912]. One study analyzed the FDG-PET of patients with localized lung cancer treated with surgery or radiation [10]. Sasaki et al. found that those with a baseline SUV of >5.0 had a significantly worse overall and disease-free survival than those with lower FDG uptake. This was true even when accounting for whichever treatment modality was used and, more importantly, for tumor stage and size as part of a multivariate analysis. The two-year overall survival was approximately 65% and 94% in those with SUV greater than and less than 5.0, respectively. This type of prognostic information gains greater importance as therapeutic options increase. For example, recent studies in lung cancer have demonstrated that adjuvant chemotherapy increases survival in those with resected disease; one could therefore envision that those with an FDG-PET SUV of less than 5.0 (who have a much better prognosis) could be spared the toxicity of such a regimen, whereas those with a high SUV would clearly be candidates for such a treatment. A prospective trial testing this approach would be timely, where subjects with a low SUV would undergo observation, whereas those with a high SUV would receive a standard or experimental adjuvant chemotherapy regimen.

Predictive Markers and Metabolic Imaging

Prognostic markers provide information about the general risks and outcomes in a group of patients with cancer. Predictive markers, on the other hand, are used to determine if a given therapy is likely to be beneficial in subsets of patients with cancer. One of the most established predictive markers in oncology is the assessment of estrogen receptor (ER) status in patients with breast cancer [13], which helps to determine the likelihood of response to antiestrogen treatment. The presence of Her2/neu in patients with breast cancer, like that of ER, provides predictive information about the response to antibody directed against this signaling pathway using trastuzumab (Herceptin). It should be noted that in each of these situations, the presence of these markers also provides prognostic information, as patients with ER or Her2/neu fare either better or worse, respectively, than patients lacking such markers. In some cases, one can use clinical information to predict patient response to a therapy. For example, patients with a common hereditary form of colon cancer (Lynch syndromes) appear to have a better prognosis without adjuvant therapy, and the presence of this genetic abnormality also predicts a lower response to 5-fluorouracil (5-FU) [14, 15]. The newer targeted agents directed against the epidermal growth factor receptor (EGFR) have greater efficacy when used in patients with mutations in this gene [16, 17]. While some of this predictive information can be obtained from the clinical setting, increasingly predictive information is being obtained from genomic and proteonomic measurements made on tumor biopsies. Such biopsies have their limitations, including the problem in which a single biopsy may not reflect the heterogeneity found in patients with multiple metastases. Measurements of gene presence by using RNA, DNA, or protein may not provide accurate information about the actual metabolic activity in a particular pathway, given the complex regulation that occurs in vivo. A recent study demonstrates the possible advantage of using metabolic imaging with PET compared to protein measurements made on biopsies, by using results in patients with ER and breast cancer. The level of ER is regularly measured by using an immunohistologic approach to obtain a semiquantitative measure of the receptor in tumor specimens. One can also image the presence of the receptor by using 16alpha-[18F]fluoro-17beta-estradiol (FES) and PET [18]. In general, the uptake of FES correlates with ER status as measured in pathologic specimens. Differences between these two approaches are seen in some patients, and may reflect differences in the ability of these approaches to predict the results of antiestrogen therapy in patients with advanced breast cancer. In general, patients with metastatic breast cancer with ER detected with immunohistology have about a 20% response to therapy with the antiestrogen tamoxifen. In a recent study, response to tamoxifen was assessed in patients with ER detected via standard immunostaining and by the functional assessment of the ER with FES PET imaging [19]. Response to tamoxifen was 24% in patients who were ER+ based on pathology and 47% for ER+ patients based on PET. Thus, functional assessment appeared superior in this pilot study, and may be worthy of routine use in patients who are slated for antiestrogen therapy. This clearly deserves further study for corroboration and to determine if PET assessment also has a predictive function in patients being treated with aromatase inhibitors.

Need for New PET Agents

At present, FDG is the only PET agent approved for routine clinical use in oncology. In fact, in the United States, FDG-PET is only approved for use in a limited number of cancers and is not approved for use in measuring treatment response, except for breast cancer. The situation varies from country to country, but in many countries the availability and use of PET is even more limited. PET can be used with a wide variety of tracers to image different aspects of tumor metabolism. Many tracers have been synthesized and tested in cell culture, a reasonable number have been studied by imaging in animals, and a few have been studied in limited trials in patients. For example, measurements of tumor energetics have also been carried out by using labeled acetate [20]. PET can be useful in measuring tumor blood flow by using 15O-water and tumor hypoxia by using tracers such as 18F-fluoromisonidazole or 64Cu-diacetyl-bis(N(4)-methylthiosemicarbazone) (ATSM) [2124].

Tumor biosynthesis can be measured with a variety of tracers. Protein synthesis has been widely studied, especially with 11C-methionine, which also measures transmethylation reactions [25]. A number of amino acids and their analogs have been developed for use of PET, such as O-(2-18F-fluoroethyl)-l-tyrosine, although none has yet found regular clinical use [26]. Membrane biosynthesis has been examined with 11C-acetate, 11C-choline, and 18F-fluorocholine [2730]. Thymidine and its analogs have been extensively explored as agents to image DNA synthesis and cell proliferation (discussed below).

Thymidine Analogs for Imaging Cell Proliferation

Thymidine labeled with tritium has found widespread use in biochemistry and cell biology for the study of DNA synthesis and cell growth. Thymine is the only nucleotide that is exclusively incorporated into DNA and not RNA, making it and its nucleoside thymidine appropriate for studying DNA metabolism. This understanding led to the synthesis of thymidine labeled with carbon-11 by Christman et al. [31] at Brookhaven. Thymidine analogs also rapidly found use in the treatment of cancer, after the initial development of 5-FU by Heidelberger et al. [32], as they can interfere with DNA synthesis. 5-FU and, subsequently, 5-fluoro-deoxyuridine were labeled with 18F for PET imaging. This allows for the in vivo pharmacokinetics and tumor uptake of both these agents. In fact, studies have demonstrated that in patients with metastatic colorectal cancer, the relative uptake of 18F-5-FU measured with PET predicts the response to treatment with therapeutic doses of the unlabeled drug. While labeled 5-FU and 5-fluoro-deoxyuridne (FUdR) may be useful in following drug kinetics, they are rapidly catabolized in the body and, therefore, less useful in imaging proliferation. 5-FU can also be incorporated into RNA, as well as DNA.

This has led to the search for other pyrimidines that can be labeled for PET and provide images of proliferation. At this point, among the most extensively studied agents are 11C-thymidine, 18F-3′-deoxy-3′-fluorothymidine (FLT), and 18F-1-(2′-deoxy-2′-fluoro-beta-d-arabinofuranosyl)-thymine (FMAU) (Fig. 2). Each of these tracers has advantages and disadvantages as a PET agent to image proliferation. Thymidine, FLT, and FMAU are all trapped in cells as part of the DNA synthetic pathway (Fig. 3). Labeled thymidine's advantage is that it is the native compound and is readily taken up by cells, phosphorylated by thymidine kinase 1 (TK), and incorporated into DNA. Pilot studies in patients have demonstrated that it can be used to produce images of tumors and response to therapy [3336]. Unfortunately, labeled thymidine is not a practical tracer for routine use in clinical studies, because it has a relatively complex synthesis and can only be labeled with 11C, which makes widespread distribution impossible. It is also rapidly degraded to labeled carbon dioxide when 11C is placed in the ring-2 position, and an even more complex set of metabolites is produced if the label is placed in the methyl position. While one can take such metabolites into account in kinetic models, this still requires extensive work by the imaging team [33, 37, 38], making routine use more difficult.
Fig. 2

Chemical structures of thymidine and its analogs. *Position where radiolabel occurs.
Fig. 3

Normal and degradation nucleoside pathways. Thymidine and FMAU are incorporated into the DNA pathway. 18F-3′-deoxy-3′-fluorothymidine (FLT) is neither degraded nor incorporated into DNA. 1-(2′-Deoxy-2′-fluoro-beta-d-arabinofuranosyl)uracil (FAU) is not degraded but is incorporated into the DNA pathway.

The placement of fluorine in deoxyribose sugar in the 2′ or 3′ positions, in FMAU and FLT, respectively, stabilizes the glycosidic bond and prevents degradation. It should be noted that glucuronidation of FLT and FMAU can still occur, and these may need to be taken into account when one wishes to dynamically model the distribution and uptake of these agents [39]. Thymidine, FLT, and FMAU are all phosphorylated by TK once taken up by the cell. Thymidine and FMAU can be trapped within the cell by DNA polymerase incorporation into DNA. FLT was originally developed as an antiviral compound, and is a DNA chain terminator and undergoes little incorporation into DNA [40]. It is retained by the action of TK as FLT phosphate. Because cells tightly control TK similar to DNA polymerase in the DNA synthetic phase of the cell cycle, the measure of retention of each of these agents reflects a measure of cell proliferation.

Initial imaging studies with FLT demonstrated that it can produce high-contrast images of tumor, such as lung cancer [41] (Fig. 4). In addition, physiologic uptake is observed in the bone marrow, a highly proliferative organ, and in the liver, kidneys, and bladder as a result of glucuronidation and clearance. FLT uptake has been found to generally correlate with proliferation of tumor, when compared to measurement of Ki-67 levels made on biopsy specimens [42, 43], but this has not been found in all situations [44]. FLT retention through phosphorylation by TK is very analogous to FDG retention via phosphorylation by hexokinase. The same three compartment models used in kinetic analysis are applicable, and one can also use the graphical analysis techniques of Patlak and Gjedde [39, 4547]. Both compartmental and graphical approaches give comparable results (r2 = 0.98). These approaches require an input function that needs either venous or arterial blood samples or estimates of such an approach from imaging the aorta or heart. While FLT undergoes little degradation, the extent of glucuronidation must be taken into account when doing kinetic analysis. Clearly, the most widespread and simplest method for quantitation of PET images is to calculate the SUV. This approach does not take into account the tracer delivery, metabolism, or kinetics, but still provides important information. This has become the standard approach to FDG-PET imaging, and is routinely used to assess tumor diagnosis, staging, and treatment response. In a series of patients with untreated breast cancer, FLT retention as measured by compartmental analysis was found to correlate with the mean SUV of the tumor (r2 = 0.85, p = 0.0002) [39]. This study was carried out in only one tumor type, and did not assess the effect of treatment. Further studies are needed to determine if measurement of SUV is as accurate in other tumor types.
Fig. 4

Whole-body image of a patient with locally advanced lung cancer. The images were obtained from a whole-body scan immediately following a 1-h dynamic scan after the injection of 360.2 mBq of [F-18]FLT. A, coronal view; B, transversal view; C, sagittal view. The dark arrows point to the tumor masses, the light arrows point to the marrow.

FLT is now being evaluated at a number of PET centers worldwide. It can be used to readily image a number of different tumor types, e.g., as lung, breast, brain, and esophageal tumors. Overall, it appears to produce images with less contrast than FDG, except in the brain [44, 4850]. Nevertheless, FLT is most likely to find use in measuring treatment response. This will require determining the best way to measure such changes and whether SUV can provide information comparable to the measurement of kinetic parameters. Further work on FLT is likely to be encouraged by the recent approval of an investigational new drug application (IND) filed by the National Cancer Institute and approved by the Food and Drug Administration.

The other pyrimidine that has been carefully evaluated is FMAU, which was originally labeled by Dr. Conti and colleagues [51] with 11C and subsequently labeled with 18F by Dr. Mangner and colleagues [52, 53]. FMAU has the potential advantage that, like thymidine, it is incorporated into DNA [51, 54]. It resists degradation, but is rapidly taken up and trapped in the liver, probably associated with glucuronidation. Its rapid liver uptake appears to decrease clearance into the kidneys and bladder, allowing for improved imaging in the pelvis compared to FLT and FDG. As a result, FMAU can be used to image primary tumors within the prostate [55]. Less FMAU retention is seen in the bone marrow, compared with that found when imaging with labeled thymidine or FLT. This may be attributable to differences in transport of the tracers, but further study is required to understand these differences and assess the impact on imaging of various tumor types. In any case, lack of marrow uptake does have the advantage in that bone metastases can be detected in patients with metastatic prostate cancer [55].

Labeled Therapeutic Compounds

The criteria for a successful therapeutic agent and successful imaging agent are very different. An imaging agent must be retained by the target, but must not affect it nor can it be too toxic for use with treatment agents. On the other hand, a therapeutic agent must reach and affect the target, but must not produce high-contrast images. Imaging labeled drugs may be useful in understanding pharmacokinetics and pharmacodynamics in tumors and normal tissues. This was previously noted in studies of labeled 5-FU, which were able to predict efficacy in patients with metastatic colon cancer. When studying labeled therapeutic agents, one often needs to label the drug itself, as even slight modifications of the molecule can greatly alter its properties. This can be seen in the vast differences, both in terms of therapy and imaging, noted with various fluorinated pyrimidine analogs. The development of labeled therapeutic agents for imaging with PET requires a great deal of effort, so one needs to determine which agents are worth the effort. Exploratory studies of H-3 and C-14 labeled compounds in animals will often provide information to determine if imaging in patients will be useful. For example, if many degradation products are rapidly generated, imaging may add little value in determining response. On the other hand, imaging the distribution of a drug may provide valuable insight into further development and choice of tumors to be treated. For example, 1-(2′-deoxy-2′-fluoro-beta-d-arabinofuranosyl)uracil (FAU) was originally developed as an antimetabolite that is incorporated into DNA and results in growth inhibition (Figs. 2 and 3). This compound was labeled for use with PET because of its potential as a therapeutic compound, but also as a possible imaging agent. It was thought that its uptake and retention might in part reflect thymidylate synthase activity, because this enzyme is required for its conversion to FMAU. Studies in dogs and patients demonstrated no retention in the marrow and little retention in colorectal cancer [54, 55]. There was visible uptake in patients with breast cancer (SUV mean = 1.3). Because this uptake is higher than the marrow (SUV 0.6), it may portend that FAU could be effective in this tumor type with decreased marrow toxicity. This hypothesis requires testing in a phase I clinical trial of the unlabeled agent.

Importance of Imaging Response in Clinical Oncology

Unfortunately, our current therapies are not usually successful for advanced cancers. If such treatments always worked, there would be no need for serial imaging to evaluate their efficacy. In addition, our treatments are often toxic and increasingly expensive, so we want to know early on if the treatment is working. If chemotherapy worked as well and was as easy as most antibiotics, there would be less need for oncologists and radiologists.

In patients with advanced cancer who are receiving routine clinical care, we want to know if therapy has failed; if so, we move on to the next treatment option. Failure, accurately measured, means we can avoid further costs and toxicity. When other therapeutic options are available, we can also move on to the next treatment choice. When developing a new therapeutic agent, investigators want to know if therapy has succeeded, and then move on to the next phase in research. Lack of success, accurately measured, means we can avoid futile treatment in patients and further development costs.

The best way to assess treatment response depends on the tumor type being studied, its stage, the treatment being utilized, and when the clinician needs to know the results. The best imaging approach and tracer may depend on all these factors. While PET imaging provides information that complements results obtained with anatomic imaging, the best timing of imaging still varies greatly, as will be noted in the following clinical examples. The most rapid responses to therapy have been observed in patients with advanced gastrointestinal stromal tumors (GIST). GIST are refractory to standard cytotoxic agents, but most of these tumors greatly overexpress the cKit receptor and are remarkably responsive to tyrosine kinase inhibitors such as imatinib (Gleevec). Studies using FDG-PET in patients being treated with imatinib have demonstrated that high FDG retention declines dramatically within a week after most successful treatments [56]. In some patients with symptoms of pain or pressure, one can also find symptomatic improvement within days of treatment, congruent with the imaging results. While one can often see improvements on PET within 24 h, the accuracy of imaging is critical if one is going to use the results to switch treatment. If waiting for 1 month improves the specificity of treatment assessment, then this would be the preferred time for repeat imaging. This is becoming an important issue as new cKit inhibitors for imatinib-resistant tumors become available.

After standard cytotoxic therapy, the best timing of therapy may be variable depending on the tumor, treatment, and accuracy needed. In patients receiving neoadjuvant treatment for pancreatic cancer, we have found that imaging after the first cycle of therapy was predictive of response and the ability to resect the tumor [57]. In patients receiving cytotoxic therapy for lymphoma, response evaluated after the first cycle of therapy may be a better predictor of disease-free survival than imaging after a number of cycles [58]. Patients with no visible FDG uptake after the first cycle had an about 80% rate of remission after 2 years compared to less than a 20% remission rate in those with tumor that was still visible. While many more patients had no tumor visible after three or more cycles of therapy, these patients often relapsed. These results indicate that the rate of improvement after treatment also provides data regarding long-term outcome of therapy. Metabolic imaging may provide information that is counterintuitive. For example, treatment of advanced breast cancer with tamoxifen can initially result in a clinical and metabolic flare of the disease, which ultimately predicts response to therapy. Clinically, the tumor flare may manifest increased bone pain, reflecting the partial agonist activity of tamoxifen that is seen before the full antagonistic effect is observed. With FDG-PET imaging, this is noted as an increase in FDG retention in those who will eventually respond [59] (Fig. 5). In fact, an increase in FDG retention 7–10 days after the start of therapy (using a 10% increase as a cutoff) resulted in a 91% positive predictive value and a 94% negative predictive value.
Fig. 5

Percent change in FDG SUV in breast cancer patients imaged 7–10 days after starting treatment with tamoxifen (N = 40). Redrawn from study by Mortimer et al. [59].

At the other end of the predictive spectrum is the use of FDG-PET to assess the response to chemoradiotherapy in patients with locally advanced lung cancer [60]. Because it can take time for the radiation to kill the cells and for inflammation to subside, imaging was performed approximately 2 to 3 months after the completion of therapy. At that time, CT was unable to predict who would be alive at 24 months. Those with tumor visible above background on PET had a 30% survival at 2 years, while those without visible tumor had an 80% survival (p = 0.0096).

Studies are beginning to assess the value of using FLT to measure therapeutic response. In mice with sarcomas treated with 5-FU, FLT uptake declined by 47.8% and 27.1% at 24 and 48 h, whereas FDG uptake decreased by only 25.6% and 33.5%, respectively [61]. For use in measuring response, FLT must provide reproducible and quantitative images. Pilot studies in patients with untreated lung cancer using FLT have demonstrated that absolute differences between images obtained about 1 week apart were less than about 20% when SUV or dynamic measurements were obtained [62]. This is similar to the results observed with FDG [63, 64]. Pilot studies in patients with breast cancer using FLT have also found significant declines in retention after the first cycle of therapy (Shields, unpublished results). Clearly, a substantial amount of work using larger numbers of subjects, different tumor types, and different treatment regimens is required to determine if FLT will find regular use in the assessment of tumor response.


FDG-PET is now a standard technique for staging of many cancer types, but it is still not optimally used or available for all types of cancers where it may be useful. Further education is needed to assist the ordering physician and the readers on how to best utilize PET. At this juncture, PET use is generally limited to the staging and restaging of cancer. The greatest unmet need is in using new imaging approaches to measure treatment response. The validity of metabolic imaging needs to be proven to the clinician before they will change therapy. Metabolic imaging may also be used in determining the efficacy of treatment as part of new drug development and licensing. Additional work is required before the regulatory agencies will accept imaging as a clinical endpoint for treatment assessment. Progress will require collaboration of the imaging community, especially in large prospective trials where PET is used in response measurements. A number of new imaging agents are undergoing testing, including FLT and FMAU, and they may find a useful place in oncology, but further studies are needed to prove their utility.

Copyright information

© Academy of Molecular Imaging 2006