¹⁸Fluorodeoxyglucose positron emission tomography (FDG-PET) has been applied increasingly to radiation treatment planning to increase the accuracy of GTV delineation besides the staging work-up. FDG-PET may be expected to visualize the “true” pathological tumor volume or reflect the “actual” tumor burden depending on the levels of glucose metabolism. Interestingly, several studies showed that PET-based GTV (PET-GTV) was often smaller than the CT- or MRI-based GTVs [29–31]. Daisne et al. showed that the macroscopic GTVs in surgical specimens were even smaller than GTVs obtained using FDG-PET, suggesting that all imaging modalities overestimate GTV. However, when examined in detail, all imaging modalities underestimated the actual tumor extent, particularly in terms of the small mucosal and submucosal infiltration of pharyngolaryngeal squamous cell carcinomas in specimens [29]. Therefore, the importance of clinical examinations for determining GTVs should be emphasized.

Unlike CT and MR images, in which the tumor has well-defined margins, it should be noted that the extent and the size of PET-GTV is highly dependent on the display windowing. Different studies have suggested various methods to determine the outline of PET-GTV, such as using the threshold of 2.5–4 of the standardized uptake value (SUV), 20–50 % of the maximal SUV, or the multiple-threshold methods [20, 31–33]. However, a standardized technique for contouring PET-GTV is still not available, and a universal standard may be impossible to establish because FDG uptakes can vary among individual tumors depending on their histological type and differentiation grade and the expression of metabolic enzymes. Moreover, PET-GTV delineation can be challenging because certain areas can be FDG avid, particularly in head and neck regions such as the tonsils and the base of the tongue, skeletal muscles, the thyroid gland, the brain, and parotid glands (Fig. 4.1a). Despite these limitations, some authors have attempted to validate the automatic segmentation of GTVs on FDG-PET using objective and user-independent methods. For example, Gregoire et al. [16] reported the use of automatically delineated PET-GTV in a dose escalation trial. In contrast, Eisbruch et al. proposed using a composite of CT-GTV and PET-GTVs [34, 35]. This proposal has been supported by several studies suggesting that local recurrences are not always detected within PET-GTVs [35, 36]. Therefore, FDG-PET should not be used as the sole modality for GTV definition, and high-risk GTV volumes could be defined based on all available information from the findings of CT, PET/CT, MRI, and physical examination.

Functional imaging using PET with various tracers could be also used to define subvolumes within GTV depending on specific biological factors. A partial dose escalation with IMRT may be attractive if we could consistently define certain subvolumes that are at a high risk of failure due to tumor hypoxia, high clonogen density, or the cell proliferation rate [37]. For example, the use of ¹⁸F-fluoromisonidazole (FMISO) PET-guided IMRT dose painting was proposed to potentially overcome the radioresistance of hypoxia in head and neck cancers [38]. However, hypoxia imaging using FMISO-PET showed significant spatial variability when repeated sessions were performed either before or during therapy; this finding suggests that reoxygenation and dynamic changes occur at the locations of tumor hypoxia [39, 40]. Thus, GTV segmentation methods using functional imaging are appealing but are still being investigated.

As described in ICRU 83, CTV is the volume of tissue that contains a demonstrable GTV and/or subclinical malignant disease with a certain probability of occurrence that is considered relevant for therapy. However, there is no general consensus regarding what probability is considered relevant for therapy. Based on clinical judgment, a possibility of microscopic disease typically over 5–10 % would be assumed to require treatment. The concepts of subclinical malignant disease suggest the possibility of both microscopic extension outside GTV and the potential involvement of regional lymph nodes, which are usually undetectable by clinical examinations or using any imaging modalities.

CTV may be created practically by adding margins (5–10 mm) to GTV, although these margins should be tailored based on anatomical and pathological considerations. Particularly, in head and neck tumors, a thorough understanding of the complex regional anatomy is required, including knowledge of the cervical compartments and fasciae that are barriers to tumor extension. Different head and neck primary tumors also exhibit a variety of spread patterns and biological behaviors. Therefore, the delineation of CTV is currently based on clinical experience and oncological considerations. Consensus guidelines and recommendations [12–18] have been proposed to help standardize the target volume selection and delineation of head and neck tumors. A 2013 update by Gregiore et al. [14] clearly illustrated several node groups not considered previously, including the supraclavicular, retroauricular, occipital, buccal, and parotid nodes. They have also shown that translation from the node levels to CTV delineation may need some adjustments. In terms of the risk of extracapsular extensions (ECEs), two previous studies reported that ECEs were limited to within 5 mm of nodes <3 cm at their largest diameter in 96 % of cases and were always within less than 10 mm [41] or that ECEs were always within 8 mm range [42]. In cases with larger lymph nodes, CTV delineation may need to consider that expansion from the nodal GTV may be limited by the surrounding structures, such as the sternocleidomastoid muscle, the paraspinal muscles, or the parotid gland. The proper implementation of these guidelines in daily practice will require common sense and good knowledge of anatomy and oncology, which could result in a reduction in treatment variations.

PTV is a geometrical concept that was introduced for treatment planning and evaluation. It is a recommended tool to shape dose distributions to ensure that the prescribed dose will actually be delivered to all parts of CTV at a clinically acceptable probability despite geometrical uncertainties such as organ motion and setup variations. The delineation of GTV and CTV is based on anatomical and oncological considerations and is independent of the irradiation technique used. In contrast, the delineation of PTV is dependent on the technique used and is part of the treatment prescription.

ICRU 83 [11] emphasized that the delineation of the primary PTV margin should not be compromised even in cases when PTV encroaches or overlaps with other PTVs, OARs, or PRVs. Developments in treatment planning software now make it possible to manage dose distributions using priority rules during optimization. Alternatively, PTV could be subdivided into regions with different dose prescriptions. Such methods may be often applicable to IMRT optimization for head and neck cancers. In some cases, PTV extends close to or even outside the patient’s skin because of tumor invasion or added margins. Most dose computation algorithms cannot accurately compute the absorbed dose in buildup regions. To overcome this limitation, ICRU 83 provided possible solutions such as PTV subdivisions and relaxation of the absorbed-dose objectives for planning.

OAR or critical normal structures are tissues that, if irradiated, could suffer significant morbidity and thus may influence treatment planning and/or the absorbed-dose prescription. In principle, all nontarget tissues could be OARs. However, normal tissues that are considered typically as OARs depend on the location of CTV and/or the prescribed absorbed dose.

From a functional perspective, OARs have been divided conceptually into “serial” and “parallel” organs [43]. In serial OARs such as the spinal cord, the integrity of each functional subunit with a serial architecture is critical to organ function, and the inactivation of any individual subunit results in dysfunction of the whole organ. In parallel OARs such as the parotid gland, independent functional subunits are organized with a parallel architecture, and it is assumed that no complication will occur until a sufficiently large number of functional subunits have been eliminated, in so-called volume effects. This concept of tissue organization is operationally useful for determining dose–volume constraints. In principle, for serial OARs showing a threshold binary response, the maximum dose in a given volume is typically the best predictor of complications. In contrast, for parallel OARs showing graded absorbed-dose responses, the mean dose or the volume that receives excessive doses of a defined value has been used to predict complications [44].

For establishing normal tissue complication probability (NTCP) models, Emami et al. reported the tolerance dose for irradiation of one-third, two-thirds, or the whole of various organs [45]. Burman et al. provided fits to Emami’s consensus dose–volume data using a Lyman model [46]. In addition, Kutcher et al. proposed a dose–volume histogram (DVH) reduction algorithm analysis that enabled the extrapolation of Emami’s constraints to any dose distribution [47]. These mathematical methods amounted to a common formula that introduced the generalized equivalent uniform dose (EUD). The resulting Lyman–Kutcher–Burman model remains the most widely used NTCP model, although it is not always the best model that should be considered [48–50]. Recently, the Qualitative Analyses of Normal Tissue Effects in the Clinic (QUANTEC) has provided a critical overview of current knowledge of the quantitative dose–response and dose–volume relationships for clinically relevant normal tissue endpoints, and QUANTEC provided practical guidelines that enabled radiation oncologists to reasonably categorize toxicity risk based on dose–volume parameters or the results of models for a number of treatment sites [51, 52]. Information provided by QUANTEC is not currently ideal, and continued adjustments and validations should be necessary after the release of further clinical data regarding dose–volume modeling [53–56].

Similar to PTV, uncertainties and variations regarding the position of OAR during treatment must be considered to avoid serious complications. For this reason, margins have to be added to OARs to compensate for these uncertainties and variations using similar principles as those used for PTV, which defines PRV.

A margin for serial OAR is much more critical than that for parallel OAR. It should be noted that the delineation of PTV and PRV often results in one or more overlapping regions. To ensure a sufficient sparing of normal tissue, the use of priority rules in the planning system for PTV or PRV can be subdivided into regions with different dose constraints. Nevertheless, ICRU recommends that the absorbed dose should be reported in the full PRV and PTV [11].

As described in ICRU 83 [11], TV is the volume of the tissue enclosed within the specific isodose envelope that receives the absorbed dose specified by the radiation oncology team as being appropriate to achieve tumor eradication or palliation, within the boundaries of acceptable complications. According to ICRU, D _{98 %} (the absorbed dose for 98 % of PTV) could be selected to determine TV. It is important to identify the shape, size, and position of TV in relation to PTV because it can provide information to evaluate the causes of local recurrences inside or outside the PTV.

In ICRU 83, RVR is defined operationally as the difference between the volume enclosed by the external contour of the patient and that of CTVs and OARs on the slices that have been imaged. RVR is important for the treatment optimization process. If RVR is not specifically determined and the dose objectives are not imposed accordingly, the planning systems for IMRT will then extensively seek any possible solutions to achieve the desired dose distribution according to the planning aims. It is possible that this could result in the delivery of unacceptably high doses elsewhere in RVR and that steep dose gradients may not be produced between PTV and RVR. To avoid these difficulties, the planning aims should be applied to RVR. In addition, the absorbed dose in RVR may be useful for estimating the risk of late adverse effects, such as carcinogenesis. Therefore, evaluating the absorbed doses in RVR is particularly important in younger patients who are expected to have a long lifespan.