PET/CT in Oncology: Current Status and Perspectives

Abstract

The discovery of the Warburg effect in the early twentieth century followed by the development of the fluorinated glucose analogue ¹⁸F-fluorodeoxyglucose (¹⁸F-FDG) and the invention of positron emission tomographs laid the foundation of clinical PET/CT. This review discusses the challenges and obstacles in clinical adoption of this technique. We then discuss advances in instrumentation, including the critically important introduction of PET/CT and current PET/CT protocols. Moreover, we provide evidence for the clinical utility of PET/CT for patient management and its potential impact on patient outcome, and address its cost and cost-effectiveness. Although this review largely focuses on ¹⁸F-FDG imaging, we also discuss a variety of additional molecular imaging approaches that can be used for cancer phenotyping with PET. Throughout this review we emphasize the critical contributions of CT to the strength of PET/CT.

Introduction

The foundation of clinical PET/CT was laid by several pivotal events (Fig. 1) that date back as far as the early twentieth century when Otto Warburg [1, 2] discovered that cancer cells switch from oxidative to glucose metabolism even in the presence of oxygen (aerobic glycolysis; Warburg effect). More than three decades later, Luis Sokoloff [3, 4] demonstrated that ¹⁴C-deoxyglucose autoradiography could be used to map and quantify functional neuroanatomical pathways ex vivo. The translation of this approach became feasible when the fluorinated glucose analogue ¹⁸F-fluorodeoxyglucose (¹⁸F-FDG) was developed [5] and Phelps and Hoffmann [6, 7] invented and built the first positron emission tomograph, which made possible the visualization of glycolytic activity in vivo.

Fig. 1.

Pivotal events in the history of PET. CMS Centers of Medicare and Medicaid Services, ¹⁴ C-DG ¹⁴C-deoxyglucose, ¹⁸ F-FDG ¹⁸F-fluorodeoxyglucose, NOPR National Oncologic PET Registry

Subsequently, ¹⁸F-FDG and ¹³N-ammonia were used to investigate regional cerebral blood flow and glucose metabolism in patients with epilepsy [8], stroke [9], cancer [10, 11], and dementia [12]. ¹⁸F-FDG PET for tumor detection was first reported in animal models in 1980 [13], and subsequently in human lung neoplasms in 1987 [14].

PET oncology research accelerated with the development of whole body PET image acquisition protocols [15]. However, clinical adoption was slow, which was explained by (1) the limited number of cyclotrons required for production of PET probes, (2) the practice patterns of oncologists that relied largely on anatomical information for determining initial and subsequent treatment strategies in cancer, and (3) limited or complete lack of reimbursement for clinical PET studies.

The hybrid PET/CT technology introduced by Townsend, Nutt, and Beyer [16] in 1998, the emergence of commercial distribution networks for ¹⁸F-FDG, together with broadened coverage by the Centers of Medicare and Medicaid Services based on an extensive literature review [17] changed the landscape of cancer imaging. More than 2,500 PET/CT scanners are currently operational in the USA, and the number of clinical PET studies exceeds two million per year.

Given the critically important role of PET/CT in oncology, an appraisal of the current status and perspectives of PET/CT is warranted. Here we discuss advances in PET/CT instrumentation and describe the information that can be derived from the combination of anatomical, functional, and molecular imaging. We then discuss current PET/CT protocols as implemented at our institution and report on initial attempts to standardize image acquisition, reconstruction, and interpretation. We also provide evidence for the clinical utility of PET/CT for patient management and its potential impact on patient outcome and address its cost and cost-effectiveness. Although this review largely focuses on ¹⁸F-FDG imaging, we will also discuss a variety of additional molecular imaging approaches that can be used for cancer phenotyping with PET. Throughout this review, we emphasize the critical contributions of CT to the strength of PET/CT.

Advances in Instrumentation

Beyer et al. reported the design of the first PET/CT scanner [16]. Its conceptual advantages included near ideal alignment between PET and CT images and CT-based corrections for photon attenuation [18]. The prototype consisted of a single-detector spiral CT scanner and a half-ring PET scanner. Initial studies conducted at the University of Pittsburgh demonstrated a diagnostic advantage of the hybrid system over PET and CT alone by more accurate lesion localization and improved diagnostic confidence [19].

Since 1998, both the PET component and the CT component of PET/CT systems have improved dramatically. For PET, fast scintillators with high stopping power such as lutetium orthosilicate and gadolinium orthosilicate have become available and have made time-of-flight PET a clinical reality [20, 21]. Routine use of time-of-flight PET led to a significant improvement in lesion detection, especially when images display significant background noise [22]. Smaller detectors resulted in improved spatial resolution. Moreover, high count rate statistics permit image acquisition times as short as 1 min per bed position in some patients [23]. Iterative image reconstruction methods such as ordered subset expectation maximization resulted in further improvements in image quality [24].

At the same time, significant improvements in CT instrumentation occurred. CT devices equipped with 64 detectors are now routinely incorporated in PET/CT. Whole-body anatomical images of high diagnostic quality can thus be acquired in a few seconds and are used for photon attenuation correction while at the same time providing diagnostic information.

Embedding Tumor Biology in Anatomy

Cancer detection, staging, restaging, and therapy monitoring has traditionally been the domain of anatomical imaging. With deeper insights into tumor biology, the limitations of the anatomical approach have become evident [25••]. For instance, soft tissue masses are composed of viable tumor, necrosis, fibrosis, and inflammation, distinctions that cannot be made on the basis of anatomical assessments. Neither the dignity of the viable tumor component nor its grade or biological behavior can be reliably predicted from the appearance, shape, or size of anatomical masses. Changes in tumor size do not reliably predict tumor responses to therapy. Histologically identical tumors may have very different genotypes and phenotypes [26], an important observation with significant therapeutic implications and consequences.

Yet anatomy provides a highly useful framework within which tumor biology can be studied with PET. Thus, PET and CT contribute equally to the value and strength of PET/CT. For instance, surgical interventions, radiation therapy, or biopsy planning rely on presurgical anatomical imaging studies. However, by more precise cancer staging, PET plays a pivotal role in stratifying patients into those who benefit versus those who do not benefit from surgery [27]. Moreover, target definition and dose painting can be improved by PET prior to radiation therapy [28] and PET can guide interventional radiologists to the most appropriate biopsy site [29].

Assessing tumor responses to therapy in cancer patients is the strength and domain of ¹⁸F-FDG PET [30]. However, combining metabolic with anatomical information by measuring total lesion glycolysis [31] or total metabolic tumor volume [32, 33] might further improve treatment response predictions.

Reading Molecular Cancer Signatures with PET

Nearly 100 years ago, Otto Warburg observed that proliferating tumor cells more readily metabolize glucose to lactate despite nonlimiting oxygen conditions. This energy-inefficient process, termed aerobic glycolysis, provides tumors with the ability to rapidly generate the macromolecules required for cell proliferation and growth. The high glucose utilization of cancer cells is enabled through a metabolic rewiring driven by altered signal transduction pathways. For example, mutations in the RAS–MAPK–ERK and PI3K–Akt–mTOR pathways and SRC can induce higher glucose uptake through modulating glucose transporter expression and translocation [34–36]. In addition to its roles in stimulating glucose transport, Akt can also promote hexokinase 1 and phosphofructokinase activity; these are two important enzymes in the production of glycolytic intermediates [37–39]. Furthermore, overexpression of the transcription factors MYC and HIFα, whose expression can be regulated by PI3K–Akt–mTOR, can influence the expression of several genes associated with glycolysis [37]. These include glucose transporters and specific enzymes that promote the aerobic glycolysis phenotype, such as PKM2 and PDK1 [40–42]. Thus, ¹⁸F-FDG PET images depict the complex interplay among gene expression [43], translation, and various signal transduction pathways (reviewed in [44]). The information extracted from these images can provide insights into tumor proliferative activity, aggressiveness [45, 46], and prognosis [47]. Moreover, ¹⁸F-FDG PET might be a useful readout of therapeutic interventions targeting one or several of these signal transduction pathways. However, a limitation is the limited specificity of ¹⁸F-FDG PET owing to physiologic glucose consumption that can occur in benign tissue (e.g., brown fat, colonic and pelvic activity, infection and inflammations, and rebound thymic hyperplasia).

Other metabolic pathways provide nutrients and metabolic building blocks for cancer cells. This is important because alternative metabolic pathways can be exploited for PET of tumors that exhibit low glycolytic activity [48]. For instance, glutamine transport and metabolism, controlled by MYC, is upregulated in many cancers [49]. This may represent an alternative to glucose metabolism, or more likely, a synergistic strategy of cancer cells to generate the energy required for growth and survival.

Increased amino acid transport and metabolism may provide important prognostic information. L-type amino acid transporter 1 (LAT1) expression was correlated with long-term outcome in lung cancer patients [50] and pancreatic cancer patients [51]. Consistently, the degree of ¹¹C-methionine uptake was correlated with LAT1 expression in glioblastoma [52], which in turn may provide prognostic information. ¹⁸F-DOPA has been used to diagnose and grade primary brain tumors [53, 54], and brain tumor recurrence [55] can be readily detected with this PET probe. Increased tumor ¹⁸F-DOPA concentration in patients with suspected brain tumor recurrence provides important prognostic information [56].

As mentioned above, the PI3K–Akt–mTOR pathway is a key regulator of tumor cell metabolism. In addition to its role in glycolysis, its activation can also promote lipid biosynthesis for cell membrane incorporation [57]. ¹¹C-choline and ¹¹C-acetate, imaging probes that target choline kinase and fatty acid synthase, respectively, have been used to image increased lipid incorporation into membrane lipid pools in primary and metastatic prostate cancer [58–63] and hepatocellular carcinoma [64]. ¹¹C-choline prostate cancer imaging has already been introduced into clinical practice in several centers in Europe [65]. Identifying the exact localization of primary, recurrent, or regionally metastatic prostate cancers and thus making possible targeted interventions requires accurate anatomical assessments, which underscores the importance of hybrid imaging modalities.

Tumor cell proliferation can be imaged with ¹⁸F-fluorothymidine (¹⁸F-FLT), a thymidine analogue that enters tumor cells via nucleoside transporters and is phosphorylated by thymidine kinase 1 [66]. It can thus serve as a marker of tumor cell proliferation. Significant correlations between ¹⁸F-FLT uptake and the expression of the proliferation marker Ki-67 have been demonstrated in lung cancer [67], colorectal cancer [68], hepatocellular carcinoma [69], and other types of cancer (reviewed in [70]). However, changes in tumor ¹⁸F-FLT uptake in response to treatment were unrelated to histopathological response and Ki-67 expression in soft tissue sarcoma [71, 72]. Thus, various chemotherapies might affect tumor ¹⁸F-FLT uptake and uncouple it from biological indices of proliferation through a variety of mechanisms [71]. The role of ¹⁸F-FLT imaging in managing cancer patients thus awaits further clarification.

Links between hormone receptor expression and ¹⁸F-FDG uptake in breast cancer have been reported. For instance, triple-negative breast cancers that are known to carry a poor prognosis exhibit significantly higher ¹⁸F-FDG uptake than estrogen and/or progesterone receptor positive tumors [73]. Thus, metabolic phenotyping of cancers with PET might provide important prognostic information.

Estrogen receptor expression can be imaged directly with PET, which permits response predictions to hormonal therapy in breast cancer [74]. Similarly, androgen receptors expressed in primary or metastatic prostate cancers have been imaged with ¹⁸F-fluorodihydrotestosterone, which binds to their ligand-binding domain [75]. More recently, a fully humanized, radiolabeled antibody against prostate-specific membrane antigen was developed to image intracellular androgen receptor signaling [76, 77].

These imaging probes might be useful for improved noninvasive phenotyping of prostate cancers, which in turn should make possible more individualized therapeutic approaches.

Other receptor-based approaches include the imaging of somatostatin and bombesin receptors in a variety of neuroendocrine tumors and prostate cancer, respectively [78]. Assessing the expression of these receptors in sometimes very small tumor lesions mandates the use of anatomical imaging for exact localization of tracer accumulation.

Highly specific imaging approaches use antibodies, diabodies, or minibodies that target cell surface structures. These molecules can be labeled with diagnostic/therapeutic radioisotope pairs, such as ⁶⁴Cu/⁶⁷Cu, ⁸⁶Y/⁹⁰Y, and ¹²⁴I/¹³¹I (reviewed in [79]). Thus, theranostic approaches have become feasible and organ tracer distribution in well-defined anatomical volumes (by CT) can provide critically important and accurate dosimetry data for radioimmunotherapy or radiopeptide therapy. The anatomical framework provided by CT is also of pivotal importance for cell trafficking by various PET reporter gene imaging approaches, for instance, when therapeutic cells are dispersed throughout the whole body [80].

In summary, a large and diverse portfolio of molecular PET probes has emerged that can be used for cancer phenotyping by addressing most of the hallmarks of cancer [81••].

Structural and Functional Information Derived from CT

Tumor size cannot be measured accurately with PET. Such measurements derived from CT (albeit with limitations) are important for T staging and for determining cancer invasion into adjacent tissues [82]. Despite considerable limitations, changes in tumor size are still most frequently used to determine tumor responses to therapy [83], and anatomical information is indispensable for the planning of biopsy, surgery, and radiation therapy.

Yet, CT images might also identify specific tumor phenotypes. For instance, tumor perfusion can be estimated by employing dynamic CT at high temporal resolution. Such measurements require the intravenous administration of contrast agents, the measurement of tissue density before, during, and after contrast agent administration, and the definition of regions of interest, from which the arterial density input function can be derived. With use of kinetic models, the density–time data of the arterial input function and those of the tumor can then be used to estimate tumor perfusion [84]. Such measurements have revealed perfusion differences between normal tissues and cancers, and tumor perfusion rates were significantly correlated with microvascular density and vascular endothelial growth factor expression in lung cancer [85, 86] and pancreatic cancer [87] (reviewed in [84]). These measurements expose patients to considerable levels of ionizing radiation, ranging from 12.3 to 36.7 mSv [88]. However, such measurements are relevant because they (1) may inform about tumor vascularity and angiogenesis and thus provide important prognostic information, (2) provide predictive information about responses to antiangiogenic drugs, and (3) may allow predictions about efficient drug delivery to tumors.

Measurements of tumor heterogeneity or tumor texture may also enhance the information that can be derived from CT images. Differences in texture might reflect variable tissue vascularization. Tumor texture was correlated with the degree of tumor hypoxia and angiogenesis [89]. Methods of texture analysis are also under development for ¹⁸F-FDG PET [90].

In summary, PET/CT is by far the most mature and comprehensive technology for structural, functional, and molecular phenotyping of cancer at the whole-body level. Its applications are near limitless in oncology and include diagnosing, staging, therapy monitoring, and treatment stratification.

PET/CT Protocols

PET/CT protocols differ among institutions and countries [91, 92]. PET/CT studies can be performed with or without single-phase or multiphase intravenous and/or oral administration of contrast agent so that fully diagnostic anatomical and molecular whole-body surveys can be obtained. Contrast CT studies for PET/CT appear to result in incremental improvements in diagnostic accuracy in some cancers. At the University of California, Los Angeles, we inject 7.4 mBq of ¹⁸F-FDG per kilogram intravenously after a 4–6 h fast (Fig. 2). Contrast agent is given orally at the time of the tracer injection. Following a 1-h uptake period, patients are positioned on the scanner table. The scan commences with a breath-hold chest CT scan to identify small lung nodules that may be missed during shallow breathing [93].

Fig. 2.

University of California, Los Angeles PET/CT protocol providing a multiphase abdominal CT scan, a breath-hold chest CT scan, and a whole-body contrast CT scan performed during shallow breathing for fusion with the PET images. IV intravenous, PO per os

Following the breath-hold chest CT scan, we intravenously administer contrast agent using protocols that best address the clinical problem. The feasibility of multiphase contrast protocols for PET/CT has recently been reported [94, 95].

The whole-body contrast CT and PET images are acquired during shallow breathing, which results in acceptable alignment between the PET and CT images [96]. The whole-body contrast CT image is used for diagnostic purposes and for attenuation correction [18]. We use oral and intravenous administration of contrast agents in all patients in whom a stand-alone CT study would employ such protocols.

We use weight-based protocols for PET studies with shorter acquisition times in light patients and longer acquisition times in heavy patients [23]. The total scan times for whole-body PET/CT protocols with intravenous administration of contrast agent average less than 30 min per patient and can be as short as 15 min. Thus, a true “one stop shop” diagnostic imaging approach has become feasible [97].

Clinical Utility of PET/CT

In 2007 we reported that ¹⁸F-FDG PET/CT was superior to conventional imaging and PET or CT alone for staging and restaging of most cancers [98]. Subsequent studies confirmed a high staging accuracy of ¹⁸F-FDG PET/CT in non-small-cell lung cancer [99], breast cancer [100–103], esophageal cancer [104, 105], colorectal cancer [106], lymphoma [107], melanoma [108], cervical cancer [109], head and neck cancers [110], bone and soft tissue sarcomas [111, 112], and myeloma [113].

A recent meta-analysis determined the accuracy of ¹⁸F-FDG PET/CT for detecting distant metastases or synchronous second cancers in more than 4,300 patients. On the basis of prospectively defined criteria, 41 published studies including patients with primary (n = 21 studies) or recurrent (n = 14 studies) cancers and patients with primary and recurrent cancers (n = 6) were included [114]. In addition, the diagnostic performance of PET/CT was compared with that of conventional imaging in more than 800 patients. Histopathology served as the gold standard in all patients. On a per patient basis, the sensitivity and specificity of PET/CT averaged 93 and 96 %, respectively, which compared favorably with the sensitivity of conventional imaging (52 %). Numerous studies have demonstrated the ability of ¹⁸F-FDG PET and PET/CT to assess tumor responses to treatment performed as early as after a single cycle of chemotherapy [115], at the middle of chemotherapy, or at the end of chemotherapy [116••, 117, 118].

¹⁸F-FDG PET/CT has also been investigated as a prognostic marker for outcome predictions. For instance, in 260 patients with Hodgkin’s lymphoma, positive PET findings after two cycles of chemotherapy were associated with a 2-year progression-free survival of 13 %, whereas 95 % of patients with negative PET findings were progression free after 2 years [119]. Other studies confirmed these reports [120]. PET-based risk-adapted therapies have also been shown to be feasible in Hodgkin’s lymphoma [121].

Similar results were reported for non-Hodgkin’s lymphoma [122] and in a variety of solid tumors [123], including cervical cancer [124], soft tissue sarcoma [115, 125] non-small-cell lung cancer [126–128], esophageal cancer [129], breast cancer [130], gastric cancer [131], and other types of cancer (reviewed in [116••, 117]).

Moreover, predictive biomarkers, i.e., those that determine whether therapeutic targets are present, have been developed. These include, among others, androgen receptor imaging [75, 132] and estrogen receptor ligands [74].

In general, tumor responses to neoadjuvant chemo(radiation) therapy correlated with the fraction of necrotic tissue in excised tumors. As a limitation, microscopic residual disease cannot be identified with PET [133].

These results have several implications. First, neoadjuvant therapy follows a predefined treatment protocol and response or nonresponse only becomes evident after surgical removal of the tumors, when changes in therapy can no longer be implemented. Thus, interim ¹⁸F-FDG PET could determine treatment efficacy early after the start of chemotherapy. Second, unnecessary toxicity of ineffective chemotherapies can be avoided. Third, postsurgical chemotherapies could be tailored on the basis of presurgical treatment responses as determined by PET.

¹⁸F-FDG PET has also been used successfully to assess tumor responses to targeted, predominantly cytostatic therapies, including imatinib [134, 135], gefitinib [136], erlotinib [137] (Fig. 3), and the B-Raf inhibitor PLX4032 [138] (Fig. 4). The prompt reduction of ¹⁸F-FDG tumor uptake in response to imatinib and gefitinib appears to be explained by a translocation of membrane-bound glucose transporters into the cytoplasm and thus their inactivation [139]. Moreover, erlotinib increases tumor tissue oxygenation and reduces the tumor uptake of PET hypoxia probes, which may also account for reductions in tumor ¹⁸F-FDG uptake in response to treatment [140].

Fig. 3.

A 54-year-old female patient with non-small-cell lung cancer and extensive metastatic disease to the bones as seen on fused sagittal images (a, arrows). Two weeks after treatment with the endothelial growth factor receptor inhibitor erlotinib, the bone metastases demonstrate near complete resolution of ¹⁸F-FDG uptake (b)

Fig. 4.

A 26-year-old female patient with metastatic melanoma with PET body maximum intensity projection images at the baseline (a) and after 8 weeks of treatment with a B-Raf inhibitor (b), demonstrating complete resolution of abnormal ¹⁸F-FDG uptake. Selected fused axial slices of a pretreatment and posttreatment hepatic lesion (c, e) and right chest wall lesion (d, f)

Combining anatomical and functional tumor response assessments may further improve treatment response assessments. Standardized uptake values (SUV) or metabolic rates in micromoles per gram per minute describe metabolic activity per gram of tissue but not metabolic rates within the entire tumor volume [141]. This limitation can be overcome with PET/CT by deriving the total lesion glycolysis (i.e., SUV × volume) [31]. These or similar approaches improved tumor response assessments in breast cancer [142] and non-small-cell lung cancer [143] but not in soft tissue sarcoma patients undergoing neoadjuvant therapy [144]. A recent review emphasizes the need for prospective studies to define the value of this integrated approach for tumor treatment response assessments [145].

Impact of PET on Patient Management

Diagnostic accuracy and the ability to assess tumor responses to treatment are only two possible end points of clinical imaging studies. Other end points include impact on patient management and outcome. The National Oncologic PET Registry, which that now includes more than 300,000 patients [146, 147], provided evidence for a highly significant impact of ¹⁸F-FDG PET on patient management across a wide variety of cancers. Patient management was affected in more than 30 % of all patients regardless of the study indication. However, the National Oncologic PET Registry did not address the impact of management changes on patient outcome. The challenges to establish such evidence have been summarized recently [148–150].

Standardization

Any imaging approach must be accurate and reproducible, should provide clinically meaningful diagnostic and prognostic information, and should improve patient management and outcome. As mentioned earlier, a high accuracy of PET/CT for diagnosing, staging, and therapy monitoring has been established. A recent meta-analysis confirmed the good reproducibility of ¹⁸F-FDG tumor uptake measurements [151].

However, PET/CT protocols are not well standardized, which is a precondition to achieve widespread adoption and acceptance of any diagnostic modality. CT-based RECIST is based on tumor size measurements whereby changes in tumor size are used to define complete response, partial response, and stable or progressive disease in response to therapy [83, 152]. Limitations of the approach have been summarized by Weber [30] and include (1) considerable interobserver variability in size measurements, (2) inaccurate differentiation between viable and nonviable tumor resulting in underestimations of responses; (3) overestimation of responses if tumor regrowth occurs rapidly, (4) inability to differentiate stable disease from beneficial response to (predominantly cytostatic) therapy, and (5) apparent stable disease that denotes slowly growing tumors rather than a beneficial response to treatment. However, although inherently inaccurate [30], RECIST provides simple guidelines for defining response or nonresponse to therapy. It is this simplicity that has led to the widespread adoption of RECIST in clinical practice.

Matters are more complicated for PET. Recent surveys highlight a substantial variability in image acquisition, reconstruction, and analysis, emphasizing the need for standardization in community-based [91] and academic [92] imaging centers.

Boellaard [153] has provided highly useful suggestions for image acquisition, reconstruction, and analysis methods. Young et al. [154] and Wahl et al. [118] have suggested PET-based treatment response criteria. Response criteria in lymphoma therapy have been adopted by many centers [155, 156]. However, no international consensus has been reached, a critically important shortcoming that needs to be addressed urgently.

Demonstrating a beneficial impact of ¹⁸F-FDG on patient outcome requires large trials with well-defined clinical end points. Over the last few years several such studies have been published. The PETAL trial was performed in lymphoma patients who had positive findings on pretreatment ¹⁸F-FDG scans [157]. Patients were initially treated with two cycles of CHOP followed by an interim PET scan. Treatment was continued in patients with negative PET findings, whereas those with positive PET findings were randomized to receive six cycles of R-CHOP versus an alternative therapy. Preliminary outcome data after 6 months revealed that relapses occurred almost six times more frequently in patients with positive findings on interim PET scans than in those with negative findings on interim PET scans (17 vs 3 %; p = 0.036).

The Municon trial was a nonrandomized study in esophageal cancer patients that used ¹⁸F-FDG PET findings 2 weeks after the start of treatment to either proceed with (in metabolic responders) or discontinue (in nonresponders) neoadjuvant therapy [158]. Nonresponding patients underwent surgery. The improved survival of metabolic responders underscored the feasibility and value of PET-guided treatment decisions. As another example, lung cancer patients who were randomized to presurgical workup with PET/CT had a significantly lower number of futile surgical procedures than those who underwent conventional staging [99]. Finally, colorectal cancer patients were randomized to either conventional follow-up or ¹⁸F-FDG PET follow-up [159]. Tumor recurrence was detected significantly earlier in the PET group, and these recurrences were more frequently cured by surgery.

Thus, PET-based risk-adapted therapy approaches are feasible and can affect patient outcome beneficially.

The Cost (Effectiveness) of PET/CT

Diagnostic tests in cancer have to meet high cost-effectiveness standards. The lack of agreement about basic principles for generating such evidence has recently been reviewed critically [150, 160].

High-end PET/CT systems range in price from $2.5 million to $3.0 million and operational costs including service contracts are substantial. In the USA, ¹⁸F-FDG PET/CT scans are currently reimbursed by Medicare at $1,150 per scan This is comparable to the reimbursement level of whole-body CT scans. Reimbursement for PET has significantly decreased (by more than 40 %) over the last 10 years. Cancer imaging expenditure accounts for approximately 4.6 % of overall Medicare cancer care costs [161]. Approximately one fifth of this, or around 1 % of Medicare expenditure, was incurred from PET/CT [162]. Although they are increasing at a lower rate, much more significant costs arise from inpatient and outpatient care, cancer drugs, physician services, and hospice care [162]. Moreover, individual procedural costs do not reflect the cost-effectiveness of diagnostic tests. Several studies have suggested that PET/CT is cost-effective across a variety of cancers by improving patient management, which in turn reduces downstream costs incurred because of incorrect management decisions [99, 163–166].

Future Perspectives

Despite the emergence of PET/MRI [167–169] the role of ¹⁸F-PET/CT in initial and subsequent patient management decisions will expand over the next decade. PET/MRI is likely to find a role in addressing some specific clinical questions. However, its high cost and operational complexity suggests that its routine clinical use will remain limited. Even currently, the use of MRI in cancer is significantly less than that of CT. For instance, in patients diagnosed with breast, colorectal, lung, and prostate cancer and lymphoma in 2006, CT was used 3.5, 12, 5.5, 4, and 6.3 times more frequently than MRI in the first 2 years after diagnosis [161]. These data are informative as they provide a realistic outlook for the potential use of or market for PET/MRI in cancer.

Concerns about radiation exposure through medical imaging have been raised [170]. Fully diagnostic PET/CT studies may expose patients to radiation doses as high as 25 mSv. However, Brenner and Hall [170] have correctly pointed to the greatly reduced relevance of this perceived risk for patients with limited life expectancy [171]. Furthermore, a recent analysis concluded that “risks of medical imaging at effective doses below 50 mSv for single procedures or 100 mSv for multiple procedures over short time periods are too low to be detectable and may be nonexistent” and that “predictions of hypothetical cancer incidence and deaths in patient populations exposed to such low doses are highly speculative and should be discouraged” because they “are harmful … and may cause some patients and parents to refuse medical imaging procedures” [172].

PET/CT applications in oncology will be refined by standardizing image acquisition, reconstruction, and analysis as well by arriving at internationally accepted treatment response criteria. Highly targeted imaging probes to determine whether therapeutic targets are expressed and active will emerge that will permit treatment stratification and thus individualized therapy approaches. Generator-based production of ⁶⁸Ga permits the labeling of peptides for peptide receptor imaging without the requirement of an on-site cyclotron. This has expanded the use of PET to include neuroendocrine tumors (⁶⁸Ga DOTATATE, ⁶⁸Ga DOTATOC) [173, 174] and for depicting neoangiogenesis using labeled RGD peptides [175]. Moreover, the labeling of bombesin receptor agonists and antagonists shows promise for imaging prostate cancer [176].

Other promising PET approaches include probes that target cell surface antigens, intracellular proteins, and hypoxia. PET reporter gene imaging will be used for trafficking of cell-based therapies. Finally, drug development will be facilitated by PET-based pharmacokinetic and pharmacodynamic studies [177]. At the same time, functional parameters such as tumor perfusion and texture will be derived from dynamic CT images that will provide indices of tumor vascularization and angiogenesis.

In summary, CT in PET/CT provides the anatomical framework within which the biology of cancer can be visualized by PET. This powerful combination will be used to further refine diagnostic, prognostic, intermediate end point, and predictive biomarkers in cancer patients. The role of PET/MRI awaits definition. We believe that in addition to the great potential in brain imaging, the advantage of reduced radiation exposure could lead to wider use in the pediatric population. Finally, cancer patients who undergo MRI studies for cancer assessments might very well benefit from the addition of PET in a single session.