early detection of treatment response, especially when targeted therapies are given, for instance in combination with radiotherapy [4–6]. Also, treatment modification to overcome radiation resistance, for instance modifying tumor cell hypoxia, may also be visualized at an earlier stage [7].
Many potential imaging biomarkers for the prediction of treatment response and treatment guidance have been described [8–12]. Before routine clinical use, these biomarkers have to bridge 2 “translational gaps”: first to become a reliable tool for medical research, and then to become a reliable, practical tool for clinical decision making [13]. Of the many tracers suggested to have potential, only a few make it into routine clinical practice. In radiotherapy, CT imaging (with or without i.v. contrast, 3D or 4D) is routinely used for treatment planning and evaluation. Magnetic resonance imaging (MRI) is increasingly being applied for target definition and treatment evaluation. Positron emission tomography (PET) is mainly used with fluorodeoxyglucose (18F-FDG), visualizing glucose metabolism as a surrogate for tumor activity, and used for diagnosis and tumor delineation. Mostly cone-beam CT or, to a lesser extent, albeit with a high potential, MRI is being applied for position verification.
Both PET and MRI can give more biological information than is currently obtained. Here, we focus on imaging biomarkers, mainly PET based, that can contribute to radiotherapy both for planning and evaluation purposes. These biomarkers should at least have the potential to allow the selection of patients benefiting from an adapted treatment (e.g., boost, hypofractionation, hypoxic cell sensitization) or exclude patients from radiotherapy that would not benefit because of radioresistance. Therefore, we will focus on 2 of the most important mechanisms of radioresistance: hypoxia and accelerated tumor cell proliferation [14] (Fig. 2).
For those who are not familiar with the technical aspects of PET, we begin by briefly explaining this technique. This is followed by the actual description of imaging hypoxia and proliferation biomarkers for radiotherapy – what do they exactly visualize and how can this be useful for radiotherapy?
Fig. 1 a. Before the change in size, other changes in tumor (cell) characteristics could be used to predict tumor development or the effect of radiotherapy. Since proliferation is an early biomarker, imaging of proliferation can be valuable for early response prediction or pretreatment patient selection. b Human tumor xenografted head and neck squamous cell carcinomas treated with a single dose of 10-Gy photons. Note the differences in rates of a changing microenvironment. Within 6–8 h a drastic reduction in hypoxia (pimonidazole, green) is observed lasting beyond the last time point at 28 h. The rate at which proliferation (BrdUrd, red) changes occur is faster and shorter; recovery of tumor cell proliferation can already be observed at 28 h. Blood vessels are blue (9F1). (Previously published by Bussink et al. [56].)
PET Imaging
PET imaging relies on the emission of a positron. Within a few millimeters of maximal emission, this (positive) positron undergoes annihilation with a (negative) electron, converting the masses into one pair of so-called annihilation photons. These 511-keV photons are emitted in about 180° opposing directions (“back-to-back”). The photons are almost simultaneously (within 6–12 ns) detected by the PET, which allows assignment of the photons to annihilation on a certain “line of response.” PET typically has an image resolution of about 5 mm, which is much larger than the scale of the biological processes being visualized.
Fig. 2. The influence of hypoxia and accelerated proliferation on the effect of radiotherapy.
Within human tissue, the half-value layer of 511-keV photons is about 7 cm. Therefore, PET images need to be corrected for attenuation. Commonly, a low-dose CT is made consecutively and used for attenuation correction. This CT is also the anatomical reference image for the PET images, to aid interpretation. Depending on the investigated tissue, it is important to understand that CT usually leads to a “snapshot” image, while PET acquisition takes at least minutes and, without a gating image, is influenced by breathing motion and heart-beating, etc. This can lead to a local mismatch between PET and CT.
PET imaging can be performed with different radionuclides, as long as positrons are being emitted. The radionuclide can be attached to a “tracer,” which is used as a vehicle and key, to lead the radionuclide to and into the tissue to be visualized. After intravenous administration, uptake and accumulation of the tracer in the target tissue takes some time. For 18F-FDG PET this is about 1 h. For hypoxia imaging this can rise up to 4 h. If not enough time is allowed every tracer is reduced to a marker of blood perfusion. Therefore, the half-life of the radionuclide should be long enough for sufficient signal at the moment of imaging. To limit the radiation dose due to the administration, the half-life should not be unnecessary long. The most used radionuclide is 18F, which has a half-life of 110 min. 18F needs to be produced in a cyclotron, but due to this half-life it can be transported after production and be used at a different location. It does not need to be produced and used on site.
The resulting PET images are quantitative, which means that voxel values correspond to the actual activity in the patient (Bq/mL). Assessment of the uptake is usually done by calculation of the “standard uptake value” (SUV) or a derivative. The SUV is the uptake in the tissue, corrected for injected activity and patient weight. In other words, if the activity was completely homogeneously distributed over the entire body, the SUV would be 1 in the whole body. If it all accumulated in half the body, the SUV would be 2 in that part and 0 in the remainder. Frequently used derivatives are SUVmax (maximum SUV in the region of interest; ROI), SUVmean (average SUV in the ROI), SUVpeak (the average SUV of a 2D or 3D ROI with a 1.2-cm diameter over the tumor area with the highest uptake [15]), MTV (metabolic tumor volume, tumor volume measured with a fixed or relative threshold, e.g., 50% of the max [15]), and TLG (total lesion glycolysis = MTV × SUVmean).
In the case of very local uptake (<2× full width of half the maximum of the system), the partial volume effect should be considered. Spillover in neighboring voxels reduces the visualized activity, and therefore reduces the SUV.
Since there are so many options in preparation, image acquisition, and image interpretation, standardization is required to enable comparison and combination of clinical study results. Therefore, the availability of guidelines of the European Association of Nuclear Medicine (EANM) [15] and their EARL accreditation program, used for the harmonization of acquisition and interpretation within multicenter trials, are also valuable in view of biomarker research.
Fig. 3. Why oxygen is a radiosensitizer: the oxygen fixation theory. a Irradiation with high energy photons or electrons
