CT remains the most widely used diagnostic imaging modality for staging and is the most frequently employed imaging modality for target delineation in cervical cancer. It is excellent in delineating lymph node regions (although its accuracy to differentiate normal from tumor-involved nodes is inferior to PET/CT). CT readily depicts the morphology of presacral target tissues and normal structures, including the small and large bowel, bladder, other abdominal organs, kidneys, and bone. CT is superior in delineating surgical clips and implanted vaginal gold markers. CT allows delineation of the uterus but does not differentiate tumor from the normal uterus, which generally is sufficient for the purposes of external beam radiation. However, substructures of the uterus, cervix, isthmus, corpus, and fundus are challenging to delineate on CT. Because of its lacking tissue differentiation between the tumor and normal soft tissue structures, CT is limited in delineating the tumor extent within the uterus. Its accuracy in parametrial involvement is inferior compared to MRI (Mitchell et al. 2006; Subak et al. 1995). Therefore, MRI is the preferred imaging modality for high-precision delineation of the primary tumor/extent as required for image-guided brachytherapy.

High-precision assessment of the primary tumor volume and extent is best assessed by MRI in cervical cancer (Bhosale et al. 2010; Balleyguier et al. 2011; Hricak 1991). With its superior soft tissue contrast, MRI is superior to CT in delineating the tumor within the cervix and differentiating it from the uterus and adjacent pelvic tissues, including bladder invasion, parametrial extension, and rectal and perirectal involvement (Mitchell et al. 2006; Subak et al. 1995). This high precision is particularly important for image-based brachytherapy, more so than in external beam radiation, where target volumes include the parauterine structures. Vaginal extension is more accurately delineated by MRI (Sala et al. 2007). MRI is superior to CT in assessing serial subtle morphologic changes in tumor volume/extent (Mayr et al. 2010), which is relevant for longitudinal imaging during therapy for adaptive therapy and for image-guided brachytherapy.

For the assessment and morphologic depiction of regional lymph nodes and postoperative seromas, CT and MRI are generally equivalent (Mitchell et al. 2006; Subak et al. 1995). For both CT and MRI, the identification of involved lymph nodes is limited to morphologic criteria, including shape and internal architecture. Configuration criteria for involvement include size (>1 cm in short axis), short-axis-to-long-axis ratio (i.e., more round shape is more consistent with involvement), and margin irregularity. Architectural criteria include central necrosis on contrast-enhanced CT (Yang et al. 2000) or MRI, signal intensity irregularity on T2-weighted MRI (Brown et al. 2003), and loss of the fatty hilum (Koh et al. 2006). With the overall limited accuracy, a combination of size, configuration, and internal architecture should be used.

MRI is helpful in identifying incidental, non-tumor-related findings that may influence therapy including distortion or retroversion of the uterus, fibroids, ovarian findings, and other pelvic abnormalities. The cervical tumor is best delineated on T2-weighted imaging, fast spin echo, or turbo spin-echo sequences, which provide high resolution, while gradient echo sequences are more susceptible to artifacts. T2-weighted sequences are superior to T1-weighted and contrast-enhanced imaging (Sironi et al. 1993). Diffusion imaging may be helpful, particularly when T2-weighted images are equivocal (Charles-Edwards et al. 2008). The use of 3 T MRI does not improve the accuracy over 1.5 T (Hori et al. 2009).

The staging and target delineation of cervical cancer have been further advanced by the introduction of fluorodeoxyglucose (¹⁸F)(FDG) PET and PET/CT. While CT and MRI identify lymph node involvement, assessment is limited to morphologic size and architectural criteria. The metabolic characterization of lymph nodes by PET/CT provides the most sensitive assessment of pelvic and para-aortic lymph node involvement.

In imaging–surgical–pathology correlation studies, CT has been reported to have an accuracy of 70–80 % for evaluation of lymph node involvement based on size. Accuracy improves to 93 % with a cutoff value of 1 cm (Hawnaur 1993; Hardesty et al. 2001; Bellomi et al. 2005).

In MRI–pathology correlation studies, MRI has been shown to provide better staging than clinical exam in early stage cervical cancer (Bhosale et al. 2010; Balleyguier et al. 2011; Hricak 1991). For tumor volume and delineation, MRI–histologic correlation was found to be 98 % in comparisons of tumor volume from 3D MRI with digitized giant histologic tissue sections (Burghardt et al. 1989; Greco et al. 1989). These studies established the high accuracy of MRI for the delineation of tumor extent in cervical cancer. In surgical–pathology correlation studies, MRI showed better sensitivities (40–57 %) and specificities (77–80 %) than CT for parametrial invasion (Siegel et al. 2012).

The sensitivity for PET ranges from 79 to 91 % (Siegel et al. 2012) with most misses related to microscopic involvement (Lin et al. 2003). PET/CT, with the added spatial resolution from CT, has a sensitivity of 58–72 %, a specificity of 93–99 %, and an accuracy of 85–99 % for lymph node involvement in cervical cancer (Choi et al. 2006; Sironi et al. 2006).

For cervical cancer, it is critical to incorporate the information from both imaging and clinical (pelvic) examination into the radiation therapy planning. Visible and/or palpable and mucosal vaginal involvement is essentially impossible to assess by axially oriented CT imaging and can also be missed on MRI. Placement of radiopaque seed markers into the cervix or into the most distal visible or palpable vaginal tumor extent assists the delineation and the target volume design. Gold markers are recommended for this purpose as ferromagnetic materials will cause metallic artifacts on MRI preventing proper imaging. A dry tampon, which provides negative contrast for CT or MRI, is a simple method to delineate the vagina and lower extent of the cervix/tumor. However, rigid vaginal markers, rectal markers, and rectal contrast administration are not recommended as they may cause upward displacement of the cervix and distension of the rectum with ensuing displacement of the cervix, respectively. If target delineation is based on such displaced anatomy, geographical tumor miss may occur during actual treatment, as daily therapy is delivered without markers/rectal distension. Vaginal tampons have been shown to not cause displacement (Weidner et al. 1999). If excessive rectal distention is noted at the time of simulation, repeat scanning after evacuation or bowel preparation regimen may increase target reproducibility. This is particularly important in postoperative radiation therapy, where mobility of the vaginal target has been well documented and overall target volumes in the pelvis tend to be smaller (Chan et al. 2008; Taylor and Powell 2008).

For CT simulation, intravenous contrast is helpful for the differentiation of regional lymph nodes from vessels, depiction of nodal architecture (see Sect. 1.3.2), and delineation of the bladder. Oral contrast given 1–1.5 h before the simulation can improve differentiation of the bowel from lymph nodes (Figs. 15.2, 15.3, 15.4, 15.5, 15.6, and 15.7), vascular structures, and tumor. However, the effect on dose calculations (Williams et al. 2002) has to be addressed during dosimetry by correcting the inhomogeneity value of the contrast-opacified bowel to water density. For MRI simulation, T2-weighted sequences to delineate the primary tumor, parametria, vagina, and lymph nodes without contrast are used, as the bladder and bowel are readily delineated on these sequences. Contrast is not needed for primary tumor delineation.

If IMRT is used, two scans, one with full and one with empty bladder, are highly recommended, particularly in postoperative patients. This allows delineation of an internal target volume (ITV) accounting for interfractional organ motion from random organ motion and variable bladder and rectal filling (Chan et al. 2008; Taylor and Powell 2008). Full/empty bladder scans are discussed in more detail in Sect. 2.4.2.

Guidelines for target delineation have been developed by the RTOG (Lim et al. 2011; Small; Small et al. 2008) and are used in RTOG and GOG studies.

The primary tumor GTV includes the cervical tumor and parametrial and vaginal tumor extensions (if MRI is used). If CT is used, which cannot differentiate tumor from the normal uterus, the entire uterus may have to be delineated as GTV, as well as CT evidence of parametrial and vaginal tumor extensions (Fig. 15.6).

The primary tumor CTV includes the cervix (if not already included in the GTV), uterus, parametria, ovaries, and a vaginal margin of at least 2 cm distal to the cervix or lowest vaginal tumor extent based on imaging evidence and per implanted markers (Fig. 15.6d). The boundaries of the parametria extend from the upper border of the fallopian tube superiorly to the pelvic floor muscles inferiorly and the posterior wall of the bladder anteriorly to the mesorectal fascia and uterosacral ligaments posteriorly. Separate delineation of the parametria is challenging on CT and is not mandatory because the PTV usually encompasses the intervening tissues between the uterus and the lymph node target at the pelvic side wall (Fig. 15.6b). If uterosacral ligaments are involved by palpation or imaging, the mesorectum and perirectal lymph nodes should be included.

In patients with known involved lymph node(s), the lymph node GTV contour encompasses the involved node on CT or MRI. Contrast CT to characterize the involved nodes by internal architectural features and/or necrosis or close consideration to architectural MRI features (Sect. 1.3.2) can be helpful. Coregistration of PET and PET/CT (even a diagnostic scan) is most useful in identifying and localizing the involved node(s). However, the delineation should be based on its morphologic geometry and margin on the CT or MRI.

The lymph node CTV contour encompasses the iliac vessels and pelvic lymph nodes, including obturator, external iliac, internal iliac, and common iliac lymph nodes (Figs. 15.2, 15.3, 15.4, 15.5, and 15.6) and 0.7 cm of the perinodal tissue for uninvolved and 1.5 cm for involved lymph nodes. Psoas and piriformis muscles at the pelvic wall and bone are excluded if the lymph nodes are uninvolved. The intraperitoneal small bowel is excluded (Figs. 15.2, 15.3, 15.4, and 15.5). Surgical clips and seromas (Mayr et al. 2011) are included where applicable. The superior extent of the lymph node CTV is typically 0.7 cm below the L4/L5 interspace (so that the PTV, which adds 0.7 cm, extends to the L4/L5 interspace).

However, the traditional 2D-based landmark of L4/L5 has been challenged as the upper border of the common iliac nodes. Significant viability of the aortic bifurcation has been found, and 40 % of common iliac lymph nodes are reported to be located above the L4/L5 interspace (Marnitz et al. 2006). Therefore, the superior border of the lymph node target should be individualized. If high common iliac lymph nodes are involved and no lymph node dissection has been performed, inclusion of the para-aortic lymph nodes should be considered. The inferior extent of the nodal target is just proximal to the inguinal canal, which represents the lower extent of the external iliac nodes. The presacral lymph nodes are included by contouring a 1–2 cm margin of tissue anterior to the sacrum from the sacral levels of S1–S3 (Figs. 15.4 and 15.5); this volume also includes the insertion of the uterosacral ligaments.

For the PTV, an additional margin of 1.5–2 cm is added to the CTV, which varies based on the degree of image guidance and modality (IMRT vs. 3D CRT) used. The distal vaginal margin adds 1 cm to the vaginal PTV, resulting in a total of 3 cm margin from the GTV. The sum of the primary tumor PTV and lymph node PTV results in a confluent final PTV that encompasses the nodal regions, tumor, uterus and parametria, uterosacral ligaments, and presacral lymph nodes (Fig. 15.6).

For the postoperative target, which is described in more detail in Sect. 2.4.2, vaginal, paravaginal, and parametrial CTVs are contoured. Delineation of the lymph node CTV follows the same principles as for intact cervical cancer.

Normal structures including the rectum (up to the rectosigmoid junction), bowel including the large bowel above the rectum, and intraperitoneal contents (small bowel and mesentery) within 5 cm above the upper border of the target volume, as well as the bladder and femoral heads, are contoured.

Endometrial carcinomas initially spread locally, traveling along the mucosa or invading into the myometrium and at times into the cervix. Direct extension into the parametria, bladder, or rectum can occur as can peritoneal spread via the fallopian tubes or through direct serosal extension. Hematogenous spread occurs primarily to the lung and liver.

A landmark surgicopathologic study was published by the Gynecologic Oncology Group and to this day remains the standard work on spread of stage I and occult stage II endometrial carcinoma (Creasman et al. 1987). It demonstrated that the likelihood of spread outside the uterus depends on a combination of factors including grade and depth of myometrial invasion. Overall, in those patients thought to have disease confined to the uterus, 22 % had extrauterine disease including 11 % with pelvic and/or para-aortic disease, 12 % with positive abdominal cytology, 5 % with adnexal involvement, and 6 % with gross peritoneal disease. It is important to note, however, that the more serious prognosis and different routes of spread seen by papillary serous and clear cell carcinoma were not recognized at the time of this study, and those tumors were not separated and evaluated independently. Unlike cervical cancer, lymphatic spread is not limited to a strictly sequential pattern from pelvic to para-aortic lymph nodes, but can occur directly to the para-aortic nodes through the lymphatics along the ovarian vessels, particularly when extensive tumor is present in the upper corpus or fundus.

Tumors with more aggressive histologies tend to metastasize to the lymph nodes and peritoneum, including the upper abdomen, more commonly, although differentiating the behavior of these tumors from aggressive grade 3 tumors is controversial (Alektiar et al. 2002).

Contemporary surgical management includes evaluation of the abdominopelvic cavity with targeted biopsy, collection of peritoneal fluid for cytology, and extrafascial hysterectomy and bilateral salpingo-oophorectomy.

Lymph node dissection is now predicated on risk of nodal spread. Distally, the dissection extends to the circumflex iliac vein and proximally to the bifurcation of the aorta into the common iliac vessels. Laterally, the dissection extends to the genitofemoral nerve that runs along the pelvic sidewall and medially to the superior vesical artery. The most posterior landmark is the obturator nerve. If the intention is to sample the para-aortic nodes, the dissection extends cephalad either to the inferior mesenteric artery (IMA) or the renal vessels. Two large randomized trials did not demonstrate an improvement in survival with surgical staging (Group 2009; Benedetti Panici et al. 2008), and only 30–40 % of patients have formal dissection in the United States (Sharma et al. 2011). The outlines of the dissection should reflect the target for irradiation if it is to substitute for nodal treatment or if added for potential residual disease postoperatively.

Familiarity of the nodal risk based on uterine risk factors such as depth of invasion, grade, and lymphovascular space invasion is essential to assist in the decision to recommend pelvic irradiation in the absence of nodal dissection. What determines an adequate level of risk is debatable, but generally the chance of serious complications from irradiation such as chronic bowel or bladder complaints or lymphedema is reported in the 2–10 % range (van Creutzberg et al. 2000; Keys et al. 2004).

Available imaging modalities have limitations in detecting spread of disease, as spread in most patients is microscopic. Local/regional staging is commonly performed utilizing CT with and without intravenous contrast and with bowel contrast. CT has been reported to have an accuracy of 70–80 % for evaluation of nodal disease based on size. Accuracy improves to 93 % with a cutoff value of 1 cm (Hawnaur 1993; Hardesty et al. 2001; Bellomi et al. 2005), and CT is the main modality to delineate lymph node regions and normal structures.

In patients treated for intact endometrial cancer, MRI is excellent in assessing the extent of myometrial and cervical invasion and extrauterine extension (Shin et al. 2011; Cade et al. 2011; Whittaker et al. 2009), which may determine treatment regimens and the decision for brachytherapy alone versus combined external plus brachytherapy. In addition, response of the tumor to external radiation can be assessed to individualize brachytherapy planning.

The (¹⁸F)PET/CT has demonstrated a 78 % sensitivity and a 93 % negative predictive value. An ongoing GOG study is evaluating the role of PET/CT in cervical and endometrial carcinoma (Signorelli et al. 2009). Magnetic resonance imaging (MRI) has demonstrated a 75–90 % accuracy rate in assessing nodal spread and invasion into the cervix or myometrium (Rockall et al. 2005; Messiou et al. 2006).

The principles of target delineation for definitive radiation to the intact uterus follow that of cervical cancer. As primary surgical management is far more common than definitive radiation for uterine cancer, this chapter will focus on largely delineation of the postoperative target.

Defining the target for postoperative pelvic radiation has been the subject of investigation and has sparked national trials (RTOG study 0418) and an atlas (http://​www.​rtog.​org/​CoreLab/​ContouringAtlase​s/​GYN.​aspx) for the postoperative treatment of cervical or uterine carcinomas (Small et al. 2008; Jhingran et al. 2012).