Protocol/image acquisition
In order to maximize the benefits of incorporating metabolic information in treatment planning and to guarantee that the images acquired comply with the requirements for treatment planning and treatment delivery, it is important to be aware of the following factors:
1.
Initial patient positioning. Accurate reproducibility of patient positioning is essential when delivering high doses to the tumor, in order to ensure tumor coverage and to protect the surrounding normal tissue, such as the rectum, small bowel, urinary bladder, and pelvic bones [
67]; therefore, immobilization devices are routinely used. Patients should be positioned in the PET/CT scanner in the treatment position using a radiation immobilization device on a flat, narrow, and rigid table top for the treatment planning, which should allow registration or indexing of immobilization devices [
67]. Immobilization systems must be individualized for each patient and should be anchored to fastening systems, which in turn must be fixed to the treatment table.
2.
Accurate alignment. Patient setup should be performed with leveling lasers with lateral and sagittal lasers, to ensure accurate alignment and positioning. The laser light system installed in the PET/CT unit must be in accordance with the one installed in the radiotherapy unit. Quality controls of the laser lights of the PET/CT system must be done routinely to maintain consistency with the treatment unit [
67] (see section VIII for quality control as well). Reference ink or tattoo marks of the isocenter should be used (one each on the right side, left side, and ventral center) to ensure reproducibility of setup at the time of treatment [
67]. Patient arms, including elbows, should be raised outside of the anticipated treatment field in a comfortable and reproducible position, usually fixed in a device above the head. In case of a sole abdominal scan, holding a ring high on the chest is an option. The [
18F]FDG PET/CT can be performed for staging and radiotherapy treatment planning in one setting. In this case, the scan may detect unexpected distant metastases, and such patients will not undergo the planned curative radiotherapy treatment, although they may have received unnecessary tattoos prior to the scan. This should be discussed with the patient before the scan. If the same [
18F]FDG PET/CT is used to perform both staging and radiotherapy treatment planning, intravenous contrast should be administered to ensure proper identification of structures, especially lymph nodes, unless contra-indicated.
3.
Combination of procedures. An alternative strategy is to perform [18F]FDG PET with low dose CT and subsequently co-register the images with a separately acquired planning CT. In this approach, it needs to be ensured that the [18F]FDG PET/CT scan still adheres to the described requirements for patient positioning and that image registration is performed with the highest possible accuracy and quality control. In case of co-registrations, registration errors can occur, so a protocol for checking these registrations should be in place.
4.
Scan region and direction. Since the pelvis is the area of focus, performing the PET scan in the caudal to cranial direction can help reduce artifacts of bladder filling and bowel peristalsis [
68]. Pelvic organs physiologically change their positions according to the fullness of the bladder, rectum, or bowels. Therefore, movement of the cervix and uterus due to bladder/bowel filling needs to be taken into account during radiation treatment planning, especially with intensity-modulated radiation (IMRT) [
69,
70].
5.
Management and evaluation of bladder filling. Bladder filling is a critical issue, as it may vary from planning (PET/)CT to treatment, and during treatment, from one fraction to another. Up to date, there is no consensus on what constitutes the best strategy to deal with this changing anatomy. The definitive imaging protocol should be developed in collaboration between departments taking all available imaging modalities into account. Limitations related to bladder filling should be considered when integrating primary staging PET/CT findings into treatment planning.
In general, priority should be given to sensitivity and specificity when the exam is performed as part of primary staging. Therefore, acquisition should be performed with an empty bladder. Patients should void just prior to the [
18F]FDG PET/CT [
71]. In most cases, this is sufficient to ensure that proper interpretation of the scan and extra intervention is not necessary. Alternatively, patients can have a Foley catheter placed prior to the [
18F]FDG injection, and then following the [
18F]FDG injection, 20–40 mg (0.5 mg/kg body weight) of furosemide can be administered intravenously along with continued normal saline aiming to give approximately 1 l of i.v. fluid [
72,
73]. It is important that the Foley catheter be placed to gravity, below the patient to allow emptying of the bladder. This approach can potentially decrease the amount of [
18F]FDG in the ureters as well as the bladder. In rare cases, it may be difficult to distinguish between local [
18F]FDG activity in ureters and small PET positive lymph nodes, and an additional limited scan after voiding could be helpful. Bladder irrigation is mainly used in diagnosing bladder cancer and is not necessary in this setting [
74].
When [
18F]FDG PET/CT is used for target volume delineation, the following options are possible:
1.
Comfortably filled bladder on the treatment planning CT and throughout the treatment. Drinking protocols are recommended to achieve this, with specifications on timing of voiding and timing and volume of fluid intake, in an attempt to have treatment as reproducible as possible [
75]. Performing PET/CT with comfortably filled bladder would be ideal for bony fusion with treatment planning CT to guide tumor target delineation, but it can be suboptimal for proper interpretation of the PET/CT images because of physiological [
18F]FDG activity in the bladder.
2.
Full and empty bladder scans at the time of treatment planning provide information about the range of internal motion of the target volumes. Performing PET/CT with empty bladder renders bony fusion hazardous but improves PET/CT interpretation by minimizing the amount of activity in the bladder. Availability of scans in both configurations provides information about the range of internal motion of the target volumes to generate an internal target volume (ITV) with individualized margins.
3.
Another approach considering bladder movements is fusion of diagnostic and treatment planning imaging series, including PET/CT, with different situations of bladder filling. These sets of scans with different anatomical situations can be used to generate a tailored ITV for the cervix and uterus region, as part of an optimization process of contouring protocols. Such complex contouring protocols based on multiple imaging series available with different combinations of bladder filling are currently being tested and evaluated prospectively in a multicenter setting in the EMBRACE II study [
76].
Interpretation/target volume delineation
The gross target volume (GTV) of the primary tumor and pathological lymph nodes are usually defined on MRI (T2 imaging), supported by gynecological examination. The metabolic tumor volume (MTV), defined as tissue with pathological [
18F]FDG uptake, is an essential part of the total volume that needs to be treated. It identifies macroscopic tumor locations, with biological characteristics that are thought to negatively affect prognosis and response to treatment and thus require inclusion in a GTV or boost area [
77]. The goal is to maximize disease control of the primary tumor and nodal metastasis alike. For this purpose, [
18F]FDG PET/CT is generally assessed using visual criteria in the appropriate clinical context. The limited spatial resolution and the “natural blurring” of the PET images mean that delineation on PET alone can be challenging. Delineation of the primary tumor and lymph nodes is primarily based on anatomical information provided by CT and/or MRI, taking into account the findings from gynecological examination, while [
18F]FDG PET/CT is mostly used for additional identification and localization of suspicious lymph nodes and detection of distant metastases.
Non-physiologic [
18F]FDG accumulation on PET images should be interpreted as pathological, especially when focal, with additional consideration of signal intensity [
61]. The identification of abnormal uptake is affected by the contrast between the tumor and its surroundings. This contrast is related to several pathophysiological factors, the most significant of which are lesion size and histology ([
18F]FDG avidity of the tumor), volume of vital tumor cells, patient movement during image acquisition, and physiological high uptake in adjacent background [
61]. This also translates to strategies to derive a contour for target definition: The border of a target volume should be positioned to enclose the metabolic tumor volume considering these factors. All available information, such as the results of the additional anatomical imaging, should be considered when defining the definitive target volume.
The primary tumor should be histologically verified before the start of any (curative) treatment. When interpreting the PET/CT in uterine cervical cancer, the histological subtype of the tumor should be taken into account (e.g., squamous cell carcinoma is highly [
18F]FDG avid, whereas mucinous adenocarcinoma often shows low [
18F]FDG uptake) [
78]. Only tumors that are sufficiently [
18F]FDG avid can be staged properly with [
18F]FDG PET/CT, and reduced sensitivity for local tumor extension and metastatic disease must be taken into account in case of less [
18F]FDG-avid tumors.
Certain standard criteria for lymph node evaluation in malignancies also apply to uterine cervical cancer [
79]. In general, lymph nodes with short axis larger than 1 cm, any lymph node with central necrosis, high IV contrast media uptake, loss of fatty hilum, or signs of extra-capsular spread should be considered pathological on anatomical images [
80]. Regardless of these criteria, corresponding [
18F]FDG activity higher than in normal surrounding tissue is suspicious for metastasis [
61]. In general, all suspicious nodes should be included in the radiotherapy treatment plan. However, some reactive nodes may also show [
18F]FDG uptake. Therefore, PET findings should be put into clinical perspective, and treatment options should be discussed in the multidisciplinary tumor board considering the known lymphatic drainage patterns in gynecological cancers: For instance, a small but [
18F]FDG-avid node in a typical nodal basin or in the vicinity of other clearly involved nodes should be considered malignant, while a similar node in an aberrant location may be ignored or considered for verification. Moreover, large nodes with massive central necrosis and only a small rim of remaining nodal tissue may show very little [
18F]FDG uptake, leading to a false-negative signal.
The diagnostic performance of PET/CT for defining pathological para-aortic lymph nodes is high with sensitivity of 83% and specificity of 91% [
25]. However, there is a possibility of false-negative para-aortic nodes on [
18F]FDG PET/CT, reported in up to 22% of those with pelvic nodal metastases [
81‐
83]. Therefore, para-aortic lymph node dissection prior to radiotherapy in patients with pelvic and without para-aortal metastases on PET/CT could be considered.
When [
18F]FDG PET/CT is used to assist in delineation of macroscopic tumor, interpretation of the images is generally visual and supported by anatomical imaging. However, visual interpretation and manual contouring of multimodal image data are subject to observer variation. Auto-contouring involves algorithm-based methods to derive tumor borders from metabolic information on PET/CT. However, this contour may not be perfect given the limited spatial resolution of PET as well as inter- and intra-tumoral biological variations and inhomogeneity. In addition, clinical MR imaging findings also need to be considered for generating an adequate GTV. As such, auto-contours require adjustments; they may assist, but cannot replace, the visual interpretation by trained observers. Modifications accounting for bladder filling status, discussed above, are also required. However, one major advantage of auto-contouring of tumor volumes may be improved interobserver agreement [
84]. There are many different auto-contouring algorithms available; it is currently unknown which performs best in the setting of cervical cancer [
85]. As previously described, a simple threshold of 40% of SUV
max can be sufficient for automatic tumor delineation on [
18F]FDG PET/CT in highly FDG-avid cervical cancer [
86], but this does not eliminate the need for subsequent visual verification and manual optimization and adjustment for clinical MR imaging findings. Especially in less [
18F]FDG-avid tumors, manual adjustment of the tumor volume is usually necessary, in particular to exclude excreted [
18F]FDG in the urinary bladder. In tumors with only mild [
18F]FDG uptake and/or small volume, the delineation of the primary tumor and involved lymph nodes on PET/CT can be challenging. In these cases, anatomical imaging, preferably MRI, should serve as the primary imaging basis for RT treatment planning.
Knowledge of pathological and physiological [
18F]FDG uptake is essential for the interpretation of PET images and in the delineation process, since several pelvic organs may show variable physiological [
18F]FDG uptake (e.g., the ovaries, endometrium, ureters, and urinary bladder) [
87]. Variable [
18F]FDG uptake in ovaries during the menstrual cycle and the differences in physiologic [
18F]FDG uptake patterns between pre- and postmenopausal women should be taken into account [
88]. Presence of distant metastases should always be discussed in the tumor board, as this is likely to change the treatment plan from curative to palliative.
In some instances, a surgical dissection of bulky nodes (mostly > 2.5 cm short axis) is performed to optimize the effect of subsequent radiotherapy [
89]. If the PET/CT is performed shortly thereafter, there is an increased chance of false-positive findings (i.e., reactive nodes) [
90]. Postsurgical lymphoceles and/or pertinent surgical clips should be included in the clinical target volume (CTV) delineation field [
91].
When the PET/CT is not acquired in the treatment position, a visual correlation between the planning CT and the PET/CT can be made, and the metabolic information can be included in the target volume delineation. Although this approach is obviously less accurate than performing the PET/CT in the treatment position, significant additional information can still be retrieved from that PET/CT compared to anatomical imaging, especially with regard to lymph node involvement [
7,
32,
92].
Visual correlation between the MRI and the PET/CT could be challenging, for example, when MR images for uterine cervical cancer are acquired perpendicularly to the long axis of the cervical canal, while the PET/CT is acquired without angulation [
8].