|Home | About | Journals | Submit | Contact Us | Français|
Protons are being used in radiation therapy because of typically better dose conformity and reduced total energy deposited in the patient as compared with photon techniques. Both aspects are related to the finite range of a proton beam. The finite range also allows advanced dose shaping. These benefits can only be fully utilized if the end of range can be predicted accurately in the patient. The prediction of the range in tissue is associated with considerable uncertainties owing to imaging, patient set-up, beam delivery, interfractional changes in patient anatomy and dose calculation. Consequently, a significant range (of the order of several millimetres) is added to the prescribed range in order to ensure tumour coverage. Thus, reducing range uncertainties would allow a reduction of the treatment volume and reduce dose to potential organs at risk.
To understand the true uncertainties and to reduce delivery errors, in vivo verification of the delivered dose or range would be highly desirable. The obvious choice is the use of in vivo dosimetry detectors that would allow real-time dose reporting.1 However, this approach is only feasible for very few indications, such as prostate cancer, where detectors could be placed on the rectal balloon used for immobilization. A more promising approach is the use of imaging methods. To use imaging devices in order to monitor treatment delivery is common practice in photon therapy where each beam penetrates the patient so that the exit dose can be measured. Protons on the other hand stop in the patient, and thus imaging can only be based on secondary radiation that is being created by the primary beam. Various methods have been proposed. For instance, MRI can be used to monitor tissue changes owing to radiation, which could potentially allow dosimetric verification.2 The downside of this method is that it cannot be used for online verification because the changes appear days or even months after treatment.
The two most promising methods for online verification are based on the fact that protons undergo nuclear interactions in tissue.3 These interactions can lead to photons that are energetic enough to penetrate the patient. For example, nuclei can be left in an excited state, which leads to high-energy (MeV) γ-radiation emitted shortly after the interaction of the primary proton with the nucleus. This radiation is thus called prompt γ-radiation. This method is very promising for the future but is currently not practical because of the lack of appropriate systems to detect these γ-rays with high efficiency and spatial resolution in a clinical setting. Thus, substantial research is still needed before this approach can find its way into clinical practice (see below).
The currently more practical approach is the use of positron emission tomography (PET) imaging. The idea to use PET for proton range verification dates back to the 1970s3,4 but was initially mentioned in pion therapy.5 Nuclear interactions of the proton beam in tissue can result in positron-emitting isotopes.6 Thus, the acquired PET images are taken without any radiotracers utilizing the activation of the patient by the treatment beam. The positron emitters of main interest are 11C and 15O with half-lives of approximately 20 and 2min, respectively. Their β+ decay leads to coincident γ-rays resulting from the annihilation of emitted positrons with electrons, which can be used to reconstruct a three dimensional distribution of positron emitters in the patient. The main purpose is to verify the range of the proton beam and not necessarily the entire dose distribution because the PET activation is not directly correlated to the dose distribution as protons lose their energy mainly via electromagnetic interactions.
Because the image reconstructed by a PET camera cannot be directly compared with the dose distribution, a different reference has to be created. For this purpose, Monte Carlo codes that are capable of not only simulating the dose but also the frequency and location of nuclear interactions are used. Such simulations create an expected PET image in the patient that can be compared (in terms of the distal signal fall-off) with the measured and reconstructed image.6 The threshold energies in the nuclear interaction cross-sections for the production of β+ isotopes are typically at a few MeV so that the Bragg peak position and the distal dose fall-off appears at a greater depth compared with the fall-off of the PET activity distribution. Nevertheless, the difference in the fall-off position between the simulated and the measured PET distribution resembles the difference in the distal dose fall-off (unless the tissue at the fall-off positions is vastly different between simulation and treatment).
Various obstacles limit the accuracy with which range can be verified with this method.7 One issue is the accuracy of the Monte Carlo simulation. The relevant cross-sections for isotope production in tissues are not known to sufficient accuracy because of (a) uncertainties in converting CT Hounsfield units into material compositions in tissue8 and (b) because experiments at physics laboratories typically focus on thin targets rather than thick targets (i.e. the patient) where a wide proton energy distribution has to be considered.9 Furthermore, the simulation results need to be corrected for biological washout because the activity in perfused tissues changes over time.10,11 Other correction factors consider that the activity is time dependent owing to the half-lives of the isotopes and that the efficiency of the detector system affects the activity distribution.8
Range verification using PET-based activity measurements has been carried out clinically in various pilot studies. Imaging can be performed either “online”,12,13 i.e. during the treatment, or “offline”,10,14,15 i.e. after the treatment has been completed. Measurements performed offline have the advantage that a PET detector is not required in the room but pose the problem of rapid decrease of the signal owing to decay and biological washout. Images taken offline predominantly show activity from radioisotopes whose half-life is longer than the time it takes to transfer the patient to the PET scanner, i.e. it reduces the signal to mainly 11C. Furthermore, patient set-up performed separately for treatment and post-treatment imaging can add an additional source of uncertainty. The main issue with online methods is the background radiation in the treatment room and the potential limited angle coverage. There is also a hybrid approach whereby a PET scanner on wheels can be moved into the treatment room shortly after the treatment has been finished (i.e. within 1–2min).16 In order to obtain an image with reasonable statistical resolution, patients have to be scanned for a few minutes which, when online methods are being used, affects the clinical workflow. The different approaches with their pros and cons have been compared in a review.17
Range verification using PET activity is currently at a cross-road. The performance shown clinically so far is typically not sufficient to suggest that a routine clinical use, both offline or online, would add significant benefit. Certainly, clinically relevant information beyond range verification might be extracted, but a range verification accuracy of 1–2mm has been shown only in favourable locations of patients with head and neck tumours close to bony structures where co-registration of the PET image with CT can be performed very accurately.10 Generally, only accuracies in range verification of the order of 3–5mm have been achieved.11
Several research approaches are currently under way to improve the accuracy of PET-based range verification. The use of PET/CT instead of PET solutions avoids errors owing to image registration but does add some imaging dose to patients. A mobile PET/CT on wheels has recently been introduced at Massachusetts General Hospital for in-room proton range verification. The addition of dual-energy CT information also has potential to improve the conversion of CT information to the material composition needed for the PET simulations.18 Furthermore, image resolution could be improved by in-beam time-of-flight PET detectors.19 Time of flight also offers the possibility of partial ring in-beam PET detectors without artefacts. In-beam detectors are able to collect data earlier, before significant radiological or biological decay. Experiments in animal systems are being conducted to improve the prediction of the biological washout that would allow a more accurate prediction of the PET signal using Monte Carlo simulations.20,21 Additional experimental data in particular for the 16O(p,3p3n)11C reaction are required to further reduce the uncertainties in Monte Carlo simulations.22
In summary, PET monitoring of proton therapy remains a promising approach and the only one currently clinically viable and actively explored by several manufacturers in new proton therapy suites. However, currently, the achievable accuracy of PET scanner-based in vivo range verification is still limited to several millimetres owing to the several issues discussed above. Nevertheless, most of these problems can be overcome in ongoing research efforts focusing on high-resolution, high-sensitivity PET that captures the biological washout following treatment. In particular, a kinetic modelling approach is being developed to derive washout-free PET activity production maps using built-in time–activity information contained in dynamic PET data.19
A possible alternative approach for in vivo range verification is using prompt γ-rays. This technique is based on the detection of prompt γ-radiation emitted after nuclear excitation by the proton beam.23,24 The latter offers distinct advantages compared with PET because of the much higher count rate at production25 and the lack of biological washout. In combination with a high-efficiency detector, the expected count rate could theoretically even allow real-time range verification. Another advantage of PET-based imaging is that the maximum in the nuclear interaction cross-sections leading to prompt γ-rays appears at a lower energy compared with the production of positron emitters. The consequence is that the prompt γ-ray fall-off is closer to the Bragg peak depth compared with the PET activity fall-off.25 There are also many disadvantages of the prompt γ-ray method, such as the absence of a coincidence signal (i.e. it is essentially a single photon emission CT method) and most importantly the absence of a suitable detector at this time. These issues, however, are to a large extent dependent on engineering aspects that can be solved. PET imaging involves biological phenomena that may be more difficult to tackle, although it can also be improved with the development of in-beam dynamic PET imaging.