Blood Vessels

The common iliac bifurcation is located approximately at the level of the third to fourth lumbar vertebrae. The common iliac arteries run anterior to the common iliac veins. The common iliac arteries are divided into internal and external iliac arteries at the level of the fifth lumbar to first sacral vertebrae. Two branches of the external iliac artery that can be identified on imaging are the inferior epigastric artery and the deep circumflex iliac artery. The internal iliac arteries course posteroinferiorly and at the lower margin of the sacroiliac joints divide into anterior and posterior trunks.¹

Pelvic Lymph Nodes

Lymphatics from pelvic organs drain to pelvic nodal chains bilaterally and to the retroperitoneum. Familiarity with the lymphatic drainage pathways is of great importance for staging pelvic tumors. Fig. 20-2 illustrates pelvic lymph node groups on axial CT images.

FIGURE 20-2 • A-D, Axial contrast-enhanced computed tomography images depicting pelvic lymphadenopathy. 1, Common iliac lymphadenopathy; 2, internal iliac lymphadenopathy; 3, external iliac lymphadenopathy; 4, obturator lymphadenopathy; 5, inguinal lymphadenopathy.

The most peripheral lymph nodes draining the pelvis are the inguinal nodes (see Fig. 20-2D). The inguinal lymph nodes are divided into superficial and deep groups. The superficial inguinal nodes drain the anus, the perianal skin, and the round ligament of the uterus. The lymph from the gluteal region and the anterior abdominal wall below the level of the umbilicus also drain to lateral nodes in this group. The medial group of superficial lymph nodes receives lymphatics from the perineal genitalia. The lower group of superficial inguinal lymph nodes receives superficial lymphatics from the lower extremity. The deep inguinal lymph nodes are located on the medial side of the femoral vein. They receive efferents from the deep lymphatics of the lower extremity, a small number of efferents from the superficial inguinal nodes, and lymphatic drainage from the glans penis or clitoris. The superficial and deep inguinal nodes drain to the external iliac lymph nodes.¹

The external iliac lymph nodes accompany the external iliac vessels and are divided into medial, posterior, and lateral groups (see Fig. 20-2B and C). The medial group of the external iliac lymph nodes drains the urinary bladder, prostate, membranous part of the urethra, cervix, and upper part of the vagina. The posterior group receives lymphatics from the internal iliac nodes via the obturator lymph nodes. The lateral group drains lymph from the superficial and deep inguinal nodes. The external iliac lymph nodes drain to the posterior and lateral common iliac nodes.¹

The internal iliac nodes accompany the internal iliac vessels and drain lymph from all of the pelvic viscera such as the body of the uterus, the prostate gland, the upper part of the vagina, the seminal vesicles, the vas deferens, the lower part of the ureters, and the bladder (see Fig. 20-2B). They also receive lymphatic drainage from the deeper parts of the perineum, the buttock muscles, and the posterior aspect of the thigh. They send efferents to the external iliac and common iliac chains. The sacral lymph nodes, which drain directly into lumbar lymphatics, and the anatomic obturator lymph nodes, which are occasionally present in the obturator canal, are members of the internal iliac group.¹

The common iliac lymph nodes accompany the common iliac vessels and are divided into lateral, medial, and posterior groups (see Fig. 20-2A). The lateral group directly drains the external iliac lymph nodes. The posterior group receives efferents from both the internal and external iliac lymph nodes. The medial common iliac group receives lymph from the internal iliac nodes. The common iliac nodes drain into the left and right lateral aortic nodal chains, which are part of the lumbar nodes.¹

The lumbar lymph nodes are divided into right lateral aortic, left lateral aortic, and preaortic lymph node groups. The right lateral aortic chain includes paracaval and retrocaval lymph nodes. The preaortic chain also includes precaval lymph nodes.¹

Bones and Muscles

The pelvic skeleton is composed of two paired innominate bones, the sacrum, and the coccyx. The innominate bones are formed by three separate bones: the ilium, ischium, and pubis. The acetabulum, which is a socket for the head of the femur, is formed by the fusion of these three bones on each side laterally. The two innominate bones unite anteriorly at the symphysis pubis and posteriorly at the sacrum.¹

The psoas muscles originate from the vertebral bodies and transverse processes of the 12th thoracic to 4th lumbar vertebrae. At the level of the first sacral vertebra, the psoas muscle unites with the iliacus muscle, which originates from the upper two thirds of the iliac fossa and sacral ala. Then, the iliopsoas muscle inserts on the lesser trochanter of the femur.¹

The levator ani and coccygeus muscles are the main muscles of the pelvic diaphragm. The levator ani muscle originates from the pubis, ischium, and the fascia of the obturator internus muscle and inserts on the coccyx and the anococcygeal raphe. The coccygeus muscle arises from the ischial spine and attaches to the coccyx posterior to the levator ani. The other pelvic floor muscles are the deep and superficial transversalis perinei muscles, the ischiocavernosus and bulbocavernosus muscles, and the external anal sphincter muscle.¹

The obturator internus muscle arising from the anterolateral wall of the true pelvis and the piriformis muscle arising from the lateral aspect of the sacrum insert on the greater trochanter of the femur and form part of the pelvis sidewall.¹

Endometrial Cancer

Endometrial carcinoma is the fourth most common cancer in females and the most common invasive gynecologic malignancy in the United States.⁶ Up to 90% of endometrial cancers are adenocarcinomas. Although the exact causal factors of endometrial cancer are still unknown, two distinct mechanisms may play a role in its origin: (1) unopposed estrogen stimulation, causing endometrial hyperplasia, which then progresses to carcinoma; and (2) spontaneous carcinoma arising from atrophic or inert endometrium.⁷

The most important prognostic factors in endometrial cancer include the histologic grade of the tumor, the stage of the tumor, depth of myometrial invasion, and the lymph node status.⁸ Because of the inability of clinical staging to assess the depth of myometrial invasion or the presence of lymphadenopathy, the International Federation of Gynecology and Obstetrics (FIGO) revised the staging of endometrial cancer to incorporate surgical findings. Surgical staging, however, is not suitable for elderly and obese patients and those with additional medical problems that make them poor candidates for surgery. Noninvasive cross-sectional imaging is particularly helpful in such cases to depict the depth of myometrial invasion, tumor extent, and presence of lymphadenopathy. Pretreatment imaging improves patient care by affecting the type and extent of surgery or radiation treatment or both.³

Ultrasound is a simple technique that can depict endometrial tumors (Fig. 20-4). The use of transabdominal ultrasonography in staging of endometrial cancer is not advocated. However, transvaginal ultrasonography is helpful in the evaluation of myometrial invasion, with accuracy rates varying between 68% and 99%.³ Myometrial invasion is suggested when the tumor disrupts the subendometrial halo and extends asymmetrically into myometrium.⁹ The limitations of ultrasound include: (1) poor soft-tissue contrast resolution, which hampers delineation between tumor, adjacent myometrium, and coexisting pathologic conditions such as uterine leiomyomas; (2) a relatively small field of view, which precludes assessment of the cervix, parametrium, and the lymph nodes, especially in patients with large tumors or large body habitus or both.³

FIGURE 20-4 • Transvaginal ultrasound depicts large heterogeneous mass containing cystic areas within the endometrial cavity.

CT is widely used for the preoperative evaluation of endometrial carcinoma to assess the extent of disease and lymph node status.³ The accuracy of preoperative staging with conventional CT for endometrial cancer varies between 84% and 88%.^10,11 However, even when helical CT was used, magnetic resonance imaging (MRI) was reported to be more sensitive and specific than CT for preoperative staging of endometrial cancer.¹²

At present, magnetic resonance is the most accurate imaging modality for the pretreatment evaluation of endometrial cancer. The reported staging accuracy of MRI for endometrial cancer is between 83% and 92%.^13–15 A suggested MRI protocol includes orthogonal T2-weighted; transverse T1-weighted; and sagittal, dynamic, contrast-enhanced, T1-weighted sequences.³ This protocol is optimal for detection of primary tumor, myometrial and cervical involvement, local spread, and lymphadenopathy.³ Current National Comprehensive Cancer Network (NCCN) guidelines do not suggest a role for fluorodeoxyglucose (FDG)-PET in the management of endometrial cancer.¹⁶ Yet FDG-PET appears highly-sensitive for the detection of metastatic endometrial cancer.¹⁷ When added to MRI or CT, FDG-PET may improve detection of malignant lymphadenopathy^18,19 and visceral²⁰ metastases, changing clinical management in approximately one third to one half of patients.^19,21 FDG-PET combined with CT appears to be more reliable than FDG-PET alone for detecting adenopathy.^18,20,22,23 Evaluation of the primary tumor by FDG-PET is limited, as FDG-PET cannot precisely determine the extent of tumor invasion, and MRI is clearly superior for that purpose. Endometrial tumors are isointense to myometrium on T1-weighted sequences and demonstrate a variable appearance on T2-weighted sequences (Fig. 20-5). Following dynamic administration of contrast, they enhance less than the normal myometrium, which is more marked on early phases.³ Stage 0 (carcinoma in situ) tumors appear as normal or widened endometrium. Stage I endometrial cancers are confined to the uterine corpus. In stage IA, the tumor is limited to endometrium and may demonstrate normal, diffuse, or focal abnormal signal intensity with a smooth endometrial-myometrial interface. Stage IB tumors penetrate into myometrium less than 50% of myometrial thickness with partial or full-thickness disruption of the junctional zone and an irregular endometrial-myometrial interface. In stage IC, the tumor extends into more than 50% of the myometrial thickness; however, the outer stripe of myometrium is intact. Stage II endometrial cancers extend beyond the uterine corpus into the cervix. In stage IIA, the internal os and endocervical canal are widened; however, the low-signal-intensity, fibrous cervical stroma is intact. In stage IIB, fibrocervical stroma is disrupted. Stage III endometrial cancers extend beyond the uterus but not to the true pelvis. In stage IIIA, the outer myometrium is irregular and disrupted. The ovaries may be involved by direct extension or as metastases. Parametrial involvement is seen as disruption of the serosa and direct tumor extension into parametrial fat. Stage IIIB tumors involve the upper vagina. Tumor involvement of the vagina is seen as disruption of the low-signal-intensity vaginal wall. In stage IIIC disease, there is lymph node involvement, which can be diagnosed when the short axis of lymph nodes exceeds 1 cm. Unfortunately, metastatic and hyperplastic lymphadenopathy cannot be distinguished based on signal intensity characteristics. Stage IV tumors extend beyond the true pelvis. In stage IVA there is disruption of tissue planes with loss of the low-signal-intensity of the bladder wall or the rectal wall, or both, indicating invasion of these organs. In stage IVB, there is distant organ metastasis.³

FIGURE 20-5 • Sagittal (A) and axial (B) T2-weighted magnetic resonance images depict endometrial cancer. Note that the low-signal-intensity junctional zone is disrupted, indicating myometrial invasion.

Cervical Cancer

Cervical cancer occurs more commonly in young women with low socioeconomic status. Risk factors include early age at first intercourse, several sexual partners, nulliparity, smoking, and a history of sexually transmitted diseases. There is strong evidence that the human papilloma virus is a main cause of cervical cancer. Cervical cancer arises at the squamocolumnar junction. The two main histologic types are squamous cell carcinoma, which accounts for 80% to 90% of cases, and adenocarcinoma, which has a worse prognosis. Tumor prognostic factors include the histologic grade of the tumor, the location of the tumor (exocervix versus endocervix), tumor volume, depth of stromal invasion, adjacent organ invasion, and lymph node status.⁵

The FIGO staging system includes findings obtained from physical examination, examination under anesthesia, cystoscopy, rectosigmoidoscopy, barium enema, biopsy, intravenous pyelography, and chest radiography. There is a discrepancy between clinical staging and intraoperative or pathologic findings in 20% to 35% of cases depending on the stage of the tumor.^5,24–26 In addition, lymph node metastases and extension into adjacent organs are difficult to detect clinically.

The diagnosis of cervical cancer is made by cytologic examination or biopsy. Imaging is warranted when the tumor diameter is greater than 2 cm, when the clinical examination to determine tumor size is difficult, or when the tumor is primarily endocervical in location. The role of imaging in the pretreatment evaluation of cervical cancer is to define the extent of tumor to decide on a suitable treatment modality.

Ultrasound has a low diagnostic value in cervical cancer staging. The patient’s habitus, lack of soft-tissue contrast, and operator dependence are limitations of ultrasound in the staging of pelvic malignancies.⁵ CT cannot differentiate normal cervical tissue from tumor and is therefore also limited for evaluating tumor size and stromal invasion⁵ (Fig. 20-6). In the staging of cervical carcinoma, CT was found to be 65% accurate and MRI was found to be 90% accurate. In evaluating lymph node metastasis, both modalities had an accuracy of 86%.²⁷ For detection of metastatic adenopathy, FDG-PET has been found more sensitive than CT²⁸ or MRI,^29,30 with equivalent or superior specificity (approximately 95% in all stages).^31,32 FDG-PET is recognized as superior to CT for the detection of abdominal paraaortic lymph node metastases in either early-stage or advanced cervical cancer.^28,31 Paraaortic lymph node metastases carry an adverse prognostic implication, necessitating detection of these sites of disease.^33,34 The superiority of FDG-PET for evaluation of paraaortic lymph node metastases has led to its use in the planning of IMRT for cervical cancer,^35,36 and FDG-PET has been used in planning doses to the primary tumor as well.^37,38 The initial clinical experience suggests that FDG-PET–guided radiotherapy improves and increases tumor irradiation without increasing doses to adjacent organs.³⁸ No large studies of FDG-PET/CT have been reported in presurgical staging of cervical cancer—only FDG-PET has been studied; FDG-PET/CT is expected to be more sensitive and specific than FDG-PET alone, as it has proven in other types of cancer.²³ Possibly, PET/CT findings may be able to limit the extent of surgical lymph node sampling, although the assay is unlikely to be sufficiently sensitive to obviate surgical lymph node staging altogether.²⁹ In a retrospective study, Grigsby and colleagues³⁹ found that patients with pretreatment FDG-PET who showed no evidence of metastases received no benefit from adjuvant chemotherapy as compared with pelvic radiotherapy alone. Whether pretreatment FDG-PET findings are a reliable basis for omitting adjuvant chemotherapy remains to be validated prospectively, in a randomized trial. FDG-PET appears very sensitive for detection of distant metastases, providing a potentially useful means of excluding patients from futile surgery.⁴⁰ The NCCN recommends FDG-PET as part of the pretreatment assessment for patients with cervical cancer of clinical stage IB2 or higher.⁴¹ In disease of a lesser stage and as part of tumor surveillance, the NCCN recommends use of FDG-PET “as clinically-indicated”⁴¹ (Fig. 20-7). In the setting of initial staging or suspected recurrence, FDG-PET (or PET/CT) changes clinical management of cervical cancer patients in approximately 21% to 55% of cases.^42–44

FIGURE 20-6 • Contrast-enhanced axial computed tomography image demonstrates large cervical mass.

FIGURE 20-7 • Fluorodeoxyglucose–positron emission tomography/computed tomography (FDG-PET/CT) of 69-year-old female with cervical cancer. Top row: Pretreatment images, including a maximum-intensity-projection image, anterior view (left); axial fusion PET/CT (middle); and axial CT (right). The pretreatment scan reveals metastatic abdominopelvic retroperitoneal adenopathy (arrowheads) and FDG uptake in the cervical lesion (white arrows). Bladder: grey arrow. Bottom row: 2 months after completion of chemoradiation (4 months after prior PET) tumor activity is no longer evident.

MRI is accepted as the most accurate imaging modality in the staging of cervical cancer and provides important information in selecting patients for surgery or radiation therapy.

Cervical cancer appears as a relatively high-signal-intensity mass within low-signal-intensity cervical stroma on T2-weighted sequences (Fig. 20-8). The contrast enhancement of cervical cancer varies considerably. Because contrast enhancement makes the tumor isointense with the surrounding cervical and parametrial tissue on T1-weighted images, it does not increase the ability of MRI to depict the tumor or stromal or parametrial invasion.^45–49 Contrast enhancement may help in the evaluation of tumor extension into the pelvic wall or adjacent organs.⁴⁵

FIGURE 20-8 • Sagittal (A) and axial (B) T2-weighted magnetic resonance images demonstrate a large cervical mass growing into the uterus.

Stage 0 (carcinoma in situ) tumors are not visible on MRI. Stage I tumors are confined to the cervix. Depending on their size, these tumors may be identified by the high signal intensity they display within a low-signal-intensity cervical ring on T2-weighted images. The depth of stromal invasion can be determined easily. Preservation of a low-signal-intensity cervical stripe is a good indicator for the exclusion of parametrial invasion. Complete disruption of the cervical ring, indicating full-thickness stromal involvement, may hamper exclusion of parametrial involvement. Stage II tumors extend beyond the cervix. In stage IIA tumors, less than two thirds of the upper vagina is involved, which is seen as loss of normal low signal intensity and thickening of the vagina. The tumor is classified as stage IIB when there is parametrial invasion. The tumor extends directly through low-signal-intensity cervical stroma into the parametrium. Stage IIIA tumors extend into the lower third of the vagina, as indicated by the loss of low signal intensity and thickening of the lower part of the vagina. In stage IIIB tumors, there is pelvic side-wall invasion, and hydronephrosis can be seen if there is ureteral involvement. Stage IV tumors invade the bladder or rectum, as indicated by the loss of normal low signal intensity and thickening of the walls of these adjacent structures.^45–49

Signal intensity characteristics are not helpful in differentiating between malignant and hyperplastic lymph nodes. MRI has a sensitivity of 62%, a specificity of 98%, and an accuracy of 93% in detecting metastatic lymphadenopathy in cervical cancer if the upper limit of normal lymph nodes is accepted as 1 cm in the short axis.⁵⁰ MRI and CT are not significantly different in the detection of metastatic lymphadenopathy in cervical cancer.⁵¹ Schwarz and colleagues⁵² prospectively studied 92 cervical cancer patients, with stages ranging from IB1 to IVA, correlating clinical outcomes (progression-free and cause-specific survival) to FDG-PET/CT findings obtained 8 to 16 weeks after completion of treatment (brachytherapy, external beam radiotherapy, and chemotherapy). The post-treatment FDG-PET/CT findings were more predictive of outcomes during the 3- to 5-year follow-up period than was the pretreatment lymph node status. At the 8- to 16-week post-therapy point, patients with a complete response by FDG-PET (i.e., a resolution of all PET evidence of metastatic disease) had a 5-year cause-specific survival rate of 84%. Patients with progressive disease detected by FDG-PET (i.e., new metastatic sites) had a survival rate of 10%. Patients with a partial PET response (i.e., no new disease, with no resolution or partial resolution of the old disease) had a survival rate of 33%. These specific results cannot be generalized to other therapeutic regimens, of course, but they highlight the fact that in patients with cervical cancer (as in patients with numerous other types of cancer), FDG-PET is effective in predicting long-term response to therapy.^53–58 The sensitivity and specificity of FDG-PET for the detection of recurrent cervical cancer appears to be similar in asymptomatic and symptomatic patients.⁵⁹ Data from more than 600 patients, cumulatively, have been reported regarding FDG-PET and the detection of recurrent cervical cancer.^{42–44,59–61} In studies with 40 or more patients,^43,44,59,61 FDG-PET has overall sensitivity ranging from 80% to100% and overall specificity ranging from 76% to 100% for the detection of recurrent cervical cancer.

Cancer of the Vagina

Most of the primary vaginal malignancies are squamous cell carcinomas. Adenocarcinomas are only 5% to 10% of primary vaginal malignancies. Clear cell adenocarcinoma of the vagina is a subtype of adenocarcinoma associated with maternal ingestion of diethylstilbestrol during pregnancy. Vaginal melanomas and vaginal sarcomas are other rare primary vaginal malignancies. In children younger than 6 years of age, the most common primary vaginal malignancy is embryonal rhabdomyosarcoma (sarcoma botryoides). Metastases to the vagina are more common than primary vaginal tumors and usually occur by direct extension from adjacent organs such as the endometrium, cervix, rectum, or bladder.⁶²

The most important prognostic factors in vaginal malignancies are tumor stage and tumor size. Staging of vaginal cancer according to FIGO includes clinical examination, chest x-ray examination, complete blood count, and the biochemical profile. Often cystoscopy, sigmoidoscopy, barium enema, and intravenous pyelography are also included in FIGO staging.⁶²

Cross-sectional imaging has been increasingly used in the evaluation of patients with vaginal cancers. Ultrasound is limited in the evaluation of the cervix and vagina. Both ultrasound and CT have low soft-tissue contrast and are therefore not good enough in depicting early vaginal cancer. CT is useful in the evaluation of advanced disease and in the assessment of lymph node status. MRI has excellent soft tissue resolution and is therefore very useful in evaluating the tumor size, location, and extent of disease.⁶²

On T2-weighted magnetic resonance images, vaginal carcinomas appear as intermediate- to high-signal-intensity masses. When the tumor is confined to the vaginal wall (stage I), the normal low-signal-intensity vaginal wall is intact on T2-weighted sequences. Stage II lesions invade paravaginal tissues but do not extend to the pelvic side wall. On T2-weighted sequences, the interface between fat and tumor is not well defined and there is medium to high signal intensity similar to tumor intensity within paravaginal tissues. In stage III tumors there is increased signal intensity within pelvic floor muscles on T2-weighted images, indicating extension into the pelvic wall. Stage IVA tumors invade the bladder or rectum or extend beyond the true pelvis. Direct tumor extension into the bladder or rectum can be seen as disruption of soft-tissue planes and increased signal intensity within the bladder or rectal wall.⁶²

Little has been written about the use of FDG-PET in vaginal cancer. Vaginal cancer is an FDG-avid disease.⁶³ FDG-PET appears to be more sensitive than conventional CT for detection of metastatic adenopathy.⁶³ In a small study⁴⁰ combining vaginal and cervical cancer patients, FDG-PET was found 100% sensitive for detection of extrapelvic metastases, thus posing FDG-PET as an attractive means for excluding patients from treatments futile in that setting (e.g., pelvic exenteration). One potential false-positive result worth noting is the presence of excreted FDG often found in the vaginal vault; small spots of contamination can mimic soft-tissue disease.

Cancer of the Vulva

Vulvar cancer is a rare malignancy occurring most commonly in women older than age 60. The most common histologic type is squamous cell carcinoma. The other vulvar cancers consist of melanoma, Bartholin gland cancer, Paget’s disease, sarcomas, basal cell carcinoma, and adenocarcinoma.⁶²

Prognosis of vulvar carcinoma depends on tumor size, depth of tumor invasion, and lymph node metastases. CT is useful in the evaluation of advanced disease and lymph node status. MRI is superior to both ultrasound and CT in evaluating local extension of the tumor.⁶²

Stage I vulvar cancers are less than 2 cm in size and can be seen as high-signal-intensity lesions on T2-weighted sequences. Stage II tumors are larger than 2 cm but confined to the vulva. In stage III tumors, there is invasion of the lower urethra with or without involvement of the vagina and anus. A high-signal-intensity mass extending into these structures can be seen on T2-weighted magnetic resonance images. In stage IVA, tumor extension to the upper urethra, bladder, rectal mucosa, or pelvic bones can be seen as areas of intermediate- to high-signal-intensity within these structures.⁶²

Although, to date, no large studies have examined the usefulness of FDG-PET in cancer of the vulva, FDG-PET can provide useful information for treatment planning and post-treatment follow-up in patients with vulvar cancer (Fig. 20-9).

FIGURE 20-9 • Fluorodeoxyglucose–positron emission tomography/computed tomography (FDG-PET/CT) of a 68-year-old woman with recurrent vulvar cancer. Top row: maximum-intensity-projection PET images, showing pretreatment staging scan (left) and follow-up scan (right), 2 months after completion of chemoradiotherapy. Although pelvic adenopathy (arrowheads, left image) resolves, after treatment the vulvar recurrence (arrow) persists and new pulmonary metastases (arrowhead, right image) developed. Axial images of the vulvar recurrence are shown before (middle row) and after (bottom row) treatment; marked metabolic activity persists in this multiple recurrent tumor.

Ovarian Cancer

Ovarian cancer is the fifth most common cause of cancer-related deaths in women and the most common cause of death from gynecologic malignancy.⁶ Approximately 90% of ovarian cancers are of epithelial origin, with subtypes such as serous, mucinous, endometrioid, clear cell, and undifferentiated. Nonepithelial cancers of the ovary include malignant granulosa cell tumor, dysgerminoma, immature teratoma, endodermal sinus tumor, and metastases.^65,66

The FIGO staging system of ovarian cancer reflects the three primary mechanisms of ovarian cancer spread: local, peritoneal, and lymphatic. Stage I ovarian cancer refers to tumor confined to one or both ovaries. Stage II indicates ovarian cancer with peritoneal metastases confined to the true pelvis. Stage III refers to ovarian cancer with extrapelvic peritoneal metastases or nodal disease. Stage IIIA indicates that there are microscopic implants. Stage IIIB refers to implants smaller than 2 cm and stage IIIC indicates that the implants are larger than 2 cm. Stage IV consists of ovarian cancer with distant metastases such as malignant pleural effusion and intrahepatic metastases. The management of ovarian cancer is closely related to stage. The standard of care for FIGO stage I, II, IIIA, and IIIB ovarian cancer is an exploratory staging laparotomy. The standard of care for resectable FIGO stage IIIC ovarian cancer is primary surgical cytoreduction (i.e., debulking) followed by adjuvant chemotherapy. Optimal debulking refers to the reduction of all tumor sites to a maximal diameter of less than 1 cm. The management of FIGO stage IV disease is primary chemo and cytoreduction. Considering different management options, the importance of imaging is in the distinction among FIGO stages IIIA, IIIB, and stage IIIC disease. Furthermore, in the category of FIGO stage IIIC disease, the importance of imaging is in the detection of nonresectable disease. The assignment of FIGO stage IV disease is also important for patient management. There are two potential pitfalls in assigning FIGO stage IV disease. On imaging, the presence of pleural effusion can be benign or malignant. The findings of pleural thickening or nodularity in addition to the presence of pleural effusion indicate the malignant nature of effusion. In the absence of those findings, thoracentesis is essential. The other important distinction is to differentiate liver surface implants (peritoneal spread and therefore stage III disease) from true intraparenchymal metastases (as a result of hematogenous spread and therefore stage IV disease).^65,67 Current NCCN guidelines suggest use of FDG-PET “as clinically-indicated” for monitoring of patients with epithelial ovarian cancer (at any stage) that has demonstrated a complete response to treatment.⁶⁸ Ovarian cancer is composed of several histologic subtypes, possibly with differing levels of FDG-avidity; however, in general, ovarian cancer is FDG-avid.^69–71 FDG-PET and FDG-PET/CT appear superior to conventional CT for staging of newly-diagnosed ovarian cancer.^69,72 For detection of recurrent disease, FDG-PET/CT and MRI appear to be grossly equivalent, overall, in sensitivity and specificity,^73,74 although MRI is superior for detection of peritoneal recurrence after cytoreductive surgery.⁷³ FDG-PET appears to be superior to CT^75–77 for the detection of recurrent disease. These generalities remain to be confirmed by well-controlled prospective trials comparing the modalities. Recurrent disease is best detected by FDG-PET in the setting of a rising cancer antigen (CA)125 value (PET sensitivities reported are approximately 85%, whereas reported PET/CT sensitivities are approximately 95%).^78–80 Only lesions smaller than 6 mm are regularly missed.^81,82 Hence, FDG-PET/CT is capable of detecting “macroscopic” disease amenable to cytoreductive surgery.⁸¹ Rising tumor marker serum levels cannot distinguish local recurrence from spreading disease, so a clear rationale for tumor localization is present in that setting. The lesion-based accuracy of FDG-PET/CT in recurrent ovarian cancer seems superior to that of PET alone.⁸³ Direct comparisons of FDG-PET/CT with CT or MRI in the setting of recurrent ovarian cancer have suffered from patient selection bias (for example, in some studies FDG-PET/CT was performed only in cases in which MRI or CT had already provided negative or equivocal results^74,78,81,84) or have used PET techniques that are no longer state-of-the-art.⁸⁵ Nevertheless, it appears that the addition of FDG-PET/CT to conventional imaging modalities frequently alters therapeutic management (>50% of cases, in some series) in patients with persistent disease or suspected recurrence, usually by detecting unresectable disease or unsuspected distant metastases.^75,78,83,86

In patients with an adnexal mass, ultrasound is very accurate in the assessment of tumor location (e.g., differentiation of uterine from adnexal masses), as well as distinguishing between a benign and a malignant adnexal lesion (Fig. 20-10). In addition to the morphologic features of the mass depicted on gray-scale ultrasound, Doppler ultrasound findings are also very helpful.^87,88 If ultrasound findings are inconclusive, MRI is recommended for further evaluation.⁸⁹ The use of contrast media in MRI allows better characterization of solid nodules within a cystic lesion or the presence of necrosis within a solid lesion. In the characterization of adnexal masses, MRI also demonstrates high inter- and intraobserver agreement.^90,91 CT is not commonly used in the characterization of adnexal masses. Characterization of an adnexal mass on CT relies on the depiction of morphologic features such as enhancing mural nodularity or heterogeneity and necrosis within the solid lesion (Fig. 20-11). The presence of ancillary findings such as ascites and peritoneal carcinomatosis are strong indicators of a malignant adnexal mass on ultrasound, CT, or MRI. For characterizing ovarian tumors, FDG-PET/CT appears at least equivalent to other modalities in sensitivity^69,71 and possibly superior to them in specificity.⁶⁹ The University of Ulm group found FDG-PET (not PET/CT) inferior.^89,92,93 Yet the Ulm reports must be reviewed critically: no reference is made to patients’ menstrual cycles. and PET/CT is generally superior to PET alone in both sensitivity and specificity, especially in the pelvis.^23,76,84 Normal ovaries can demonstrate prominent FDG uptake that is physiologic and variable,⁹⁴ related to the menstrual cycle;^94,95 hence, premenopausal patients with an adnexal mass must be scheduled for FDG-PET at an appropriate day in the cycle to avoid false-positive results; in general, the week before and after midcycle ought to be avoided.

FIGURE 20-10 • Ovarian serous adenocarcinoma. Transvaginal ultrasound depicts a cystic ovarian mass with nodular solid components.

FIGURE 20-11 • Contrast-enhanced axial computed tomography reveals a left ovarian cystic mass with thick nodular wall.

Ovarian cancer can spread by local continuity to the opposite ovary; uterus; and adjacent structures such as the sigmoid colon, urinary bladder, and pelvic sidewall. Imaging criteria for rectosigmoid and urinary bladder invasion include focal obliteration of the fat plane between the tumor and these organs.

Peritoneal spread is the most common route of ovarian cancer spread. Peritoneal nodules are commonly seen in the pouch of Douglas, right paracolic gutter, right subphrenic space, and left paracolic gutter. Peritoneal metastases appear as nodular or plaquelike enhancing soft-tissue masses of varying sizes. Ascites in a patient with ovarian tumor usually indicates its malignant nature. Ascitic fluid often outlines small peritoneal implants, facilitating their detection.

Ovarian cancer can also spread to lymph nodes. Most commonly, nodal disease is seen along the aortocaval and paraortic retroperitoneal sites. Second, nodal spread along the broad ligament can reach the internal iliac, obturator, and external iliac vessels. Third, lymph node involvement can occur via the round ligament to the external iliac and inguinal nodes. Superior diaphragmatic lymphadenopathy is considered distant nodal disease and indicates advanced ovarian cancer. Extra-abdominal lymph node metastases, at presentation, have traditionally been thought to be rare in ovarian cancer based on clinical examination. Yet using FDG-PET/CT, Risum and colleagues⁹⁶ found extra-abdominal nodal metastases in 37% of patients with newly diagnosed ovarian cancer.

Hematogeneous spread from ovarian cancer commonly occurs to sites such as the liver, lung, pleura, adrenal gland, and spleen. Bone and brain metastases are rare.

FDG-PET has demonstrated a considerable ability to predict and monitor the response of ovarian cancer to therapy. For example, detection of large-bowel mesenteric implants by pretreatment FDG-PET/CT was an independent predictor of incomplete cytoreduction in a multivariate analysis (primarily of patients with stage IIIC disease).⁹⁶ In a study of patients with stage IIIC and IV disease, reductions in tumor FDG uptake, induced by presurgical carboplatin-based chemotherapy, were predictive of significantly prolonged survival, whereas clinical response data (including CA125 response) and histopathologic tumor response data were not.⁹⁷

Prostate Cancer

Prostate cancer is the most common cancer seen in men and the second most common cause of cancer-related deaths after lung cancer. It is estimated that 186,320 new cases of prostate cancer occurred among men in the United States in 2008. However, 91% of these new cases were expected to be diagnosed at local and regional stages, which have 5-year relative survival rates of nearly 100%.⁶ Autopsy studies showed that 30% to 46% of men older than age 50 have microscopic prostate cancer, but less than 20% of men develop clinically overt prostate cancer in their lifetime.^103,104 Prostate cancer has less predictable behavior than do other aggressive cancers. Prognostic factors include the size, the histologic grade, the stage of the tumor, and age.

Prostate cancer is suspected because of an abnormal digital rectal examination or elevated serum prostate-specific antigen (PSA) level, which are generally recommended in yearly check-ups for men older than age 50. Diagnosis of prostate cancer is made by histopathologic examination of tissue specimens obtained from the prostate, which also allows estimation of tumor grade. Imaging findings have no role in the early detection and diagnosis of prostate cancer, but provide valuable information in staging the disease. Another role of imaging is guiding biopsy.¹⁰⁰

Transrectal ultrasound provides high-resolution images of the prostate gland because a high-frequency (5- to 7.5-MHz) probe can be placed close to the prostate. Ultrasound also allows examination of the prostate in multiple planes and provides very accurate measurements of the prostate gland. The other major use of transrectal ultrasound is guidance for biopsy.¹⁰⁰ Only prostate cancers located in the peripheral zone can be reliably detected by ultrasound. Prostate cancer is usually seen as a hypoechoic area relative to the peripheral zone (Fig. 20-13). However, up to 40% of prostate cancers are isoechoic, hampering their detection with ultrasound.^105,106 The finding of a hypoechoic area within the peripheral zone is not specific for prostate carcinoma and can also be seen in benign processes such as prostatitis.^100,105 Distortion of the normal internal anatomy of the prostate is another important sign that may help in detection of prostate carcinoma, especially in advanced cases.¹⁰⁵

FIGURE 20-13 • Transrectal ultrasound of the prostate depicts hypoechoic tumor within the peripheral zone.

CT, even with contrast enhancement, lacks the soft tissue resolution for the detection of prostate cancer within a normal prostate. Additionally, CT has no advantages over ultrasound in providing guidance for biopsy.¹⁰⁰ However, CT may be useful in the detection of distant metastases and lymph nodes.¹⁰⁰

MRI provides the best images of the normal zonal anatomy of the prostate.¹⁰⁰ On T2-weighted magnetic resonance images, prostate cancer is seen most commonly as a low-signal-intensity area within the high-signal-intensity normal peripheral zone. The detection of prostate cancer on MRI, as on transrectal ultrasound, may be hampered by postbiopsy changes, prostatitis, dystrophic changes, or changes secondary to radiation; or hormonal ablation therapy can be seen as low-signal-intensity areas mimicking tumor.¹⁰⁰ It has been shown that MRI should be performed at least 3 weeks after biopsy to prevent under- or over-staging of tumor extent on MRI. The role of MRI in evaluating prostate cancer is mainly to demonstrate extracapsular and seminal vesicle invasion (Fig. 20-14). MRI findings of extracapsular extension include irregular bulge of the prostate margin, contour deformity with step-off or angulated margin, overt disruption of the capsule with direct tumor extension, obliteration of the recto-prostatic angle, and asymmetry of neurovascular bundles. Seminal vesicle invasion is diagnosed when contiguous low-signal-intensity tumor extension into and around seminal vesicles is demonstrated, or when tumor extension along the ejaculatory duct results in nonvisualization of the ejaculatory duct, decreased signal intensity of seminal vesicles, and loss of the seminal vesicle wall on T2-weighted images.^107,108

FIGURE 20-14 • Axial (A) and coronal (B) T2-weighted magnetic resonance images of the prostate reveal a large tumor involving the right side of prostate. The tumor invades the seminal vesicles (B).

Magnetic resonance spectroscopy is a technique used in prostate imaging that provides metabolic information by detection of cellular metabolites: citrate, creatine, and choline. In tumor localization, combined use of MRI and magnetic resonance spectroscopy provides the highest specificity (91%) among noninvasive methods. Combined use of MRI and magnetic resonance spectroscopy also improves the detection of extracapsular extension.^109–111

Current NCCN and American College of Radiology guidelines do not include FDG-PET in the management of prostate cancer.¹¹² Initially, FDG-PET was thought to be useless in prostate cancer, as FDG-PET scans of prostate cancer patients frequently failed to detect sites of viable tumor.^113,114 Schöder and colleagues¹¹⁵ provided evidence that PSA values and PSA kinetics can be used to determine the likelihood that FDG-PET imaging will detect tumors in the setting of PSA relapse; the authors suggested threshold values of PSA 2.4 ng/mL and PSA velocity 1.3 ng/mL/year for predicting an abnormal FDG-PET scan. Seltzer and colleagues¹¹⁶ found similar findings in a smaller group of patients. In castrate patients with progressive disease, prostate cancer is most often FDG-avid.⁵⁴ In prostate cancer patients with a high likelihood of FDG-avid disease, FDG-PET has a role in distinguishing local from distant recurrence and serves as a marker of tumor response to therapy. Tumor FDG-avidity portends a worse clinical outcome for the prostate cancer patient in terms of relapse-free survival.¹¹⁷ This phenomenon has been observed in thyroid cancer,¹¹⁸ leukemia, and lymphoma¹¹⁹ wherein “hypermetabolism” of glucose, visualized by FDG-PET, appears associated with an aggressive tumor phenotype. Hence, lack of FDG uptake, by tumors—although frustrating from a diagnostic perspective—can be reassuring from a prognostic perspective, even when disease is known to be present. Shreve and colleagues noted an apparent association between high FDG-avidity and hormone-refractory disease.¹¹⁴ Animal models demonstrate that tumor FDG-avidity correlates with changes in systemic hormone levels.^120,121

Evaluation of the prostate gland is often limited by the presence of excreted FDG in the bladder; some centers have taken a logical approach of catheterizing and draining the bladder before imaging.^{113,117,122–124} The primary tumor of untreated prostate cancer is usually FDG-avid¹¹⁷ and may be detected as an incidental finding.¹²⁵ FDG uptake in a prostate lesion does not distinguish malignant from benign disease (e.g., BPH 113 or 123, or focal prostatitis). Nor can FDG-PET differentiate local recurrence from scar tissue after radical prostatectomy, according to one study.¹²²

FDG-PET appears sensitive for metastatic pelvic adenopathy if it is not a subcentimeter lesion.^114,124,126 In the detection of metastatic pelvic adenopathy from prostate cancer, as in the detection of other pelvic cancers, FDG-PET/CT is expected to have greater sensitivity and specificity than FDG-PET.²³ In early studies, bladder radioactivity sometimes obscured evaluation of pelvic lymph nodes, but this is less of a problem with contemporary data-processing algorithms.¹¹⁴

FDG-PET is reportedly less sensitive for osseous metastases than is bone scintigraphy (e.g., ^99mTc-methylene diphosphonate or ^99mTc-hydroxymethylene diphosphonate),¹¹⁴ yet this finding is controversial.^54,127 The reliability of bone scintigraphy and CT as assays of skeletal tumor burden is generally recognized as poor: metastases centered in the marrow of a bone often go undetected; and these marrow metastases, when dying of anticancer therapy, can appear as “new” lesions on CT and bone scintigraphy (the so-called “flare response phenomenon”). Therapy-induced changes in tumor FDG uptake are concordant with standard clinical measures in approximately 95% of cases⁵⁴; in assaying therapy response, FDG-PET likely has a role as a substitute for PSA in patients whose tumors produce scanty levels of PSA; to verify the absence of an emerging non-PSA–producing tumor phenotype; and as an arbiter when PSA and conventional imaging are discordant.⁵⁴

NCCN guidelines suggest scintigraphy with the radiolabeled antibody capromab pendetide (ProstaScint) as an option for diagnostic imaging in the salvage work-up of prostate cancer patients.¹¹² This Food and Drug Administration–approved radioimmunoscintigraphy assay detects glutamate carboxypeptidase II, also known as prostate-specific membrane antigen (PSMA), on its intracellular domain, which in vivo is usually exposed only by dead or dying prostate cells. Early clinical studies suggested PSMA-immunoscintigraphy was superior to CT and MRI for disease staging. After its wider dissemination into routine clinical settings, however, enthusiasm seems to have waned considerably; the reliability of PSMA scanning for tumor detection in routine clinical settings did not appear to live up to the expectations of some. The added value of radioimmunoscintigraphy to other available imaging modalities appears questionable.¹¹⁶ Because capromab pendetide is a murine antibody, it is potentially immunogenic, although adverse reactions occur in fewer than 4% of patients and are usually mild.

Yet PSMA-immunoscintigraphy should not be abandoned prematurely; two recent advances promise to improve PSMA imaging: (1) single photon emission computed tomography (SPECT)-CT and (2) novel PSMA probes. SPECT-CT is a combination of scintigraphic 3-D imaging with conventional x-ray CT. SPECT-CT has recently begun a wider dissemination into the routine clinical setting, with multiple commercial vendors now offering scanners incorporating gamma cameras and an x-ray CT into one system. These “in-line,” integrated, hybrid systems provide CT-based anatomic localization of scintigraphic signals—so-called “fusion imaging”—which has proven extremely useful for distinguishing true foci of disease from artifacts or false-positive foci. In addition, the CT data provides attenuation-correction that improves the quality of SPECT images, as do modern SPECT image reconstruction software algorithms. Software-based “fusion” or “coregistration” of CT (or MRI) and SPECT image datasets acquired from separate scanners is an alternative. Yet software-based fusion is generally considered less reliable than hardware-based fusion.

New PSMA probes, such as the antibody hJ591, detect the extracellular domain of PSMA expressed by viable prostate cells, which may improve the sensitivity of radioimmunoscintigraphy. As a humanized antibody, hJ591 appears less immunogenic than the 7E11 antibody on which capromab pendetide is based. The hJ591 antibody is currently radiolabeled for SPECT imaging, but an hJ591 radioantibody for PET imaging is also in development.

Other PET tracers are in clinical testing as agents for detection of prostate tumors. In current literature, leading contenders include radiolabeled analogs of testosterone, choline, and acetate. These tracers are experimental and not yet widely available, but promise to overcome some of the limitations of FDG. The main prostatic form of androgen, 5α-dihydrotestosterone, is an analog of 16β-[¹⁸F]fluoro-5α-dihydrotestosterone (FDHT). FDHT binds the androgen receptor. Ectopic FDHT uptake, visualized by PET, promises to be a more prostate-specific manifestation of metastases than is offered by other commonly used imaging characteristics (e.g., abnormal osseous activity on bone scans, sclerotic lesions on CT, and patterns of MRI signals).

Radiolabeled analogs of choline and acetate have generated interest in prostate cancer as clinical PET trials indicate these agents can be more sensitive than FDG for detection of prostate cancer metastases. Like FDG, these tracers are not prostate-specific, but localize to disease (including prostatic tumors) on the basis of increased metabolism.

The tumor-node-metastasis (TNM) staging classification is widely used for staging prostate cancer and can be applied to imaging. T₁ prostate cancers are nonpalpable tumors that are discovered incidentally during transurethral resection of the prostate or because of an elevated PSA. T₂ tumors represent clinically localized palpable cancers. The tumor is classified as stage T₃ if it spreads beyond the prostatic capsule. T₄ tumors represent cancers invading adjacent structures such as the bladder, rectum, or pelvic side wall. The N classification denotes the presence or absence of lymphatic involvement, and the M classification indicates distant metastasis.

The primary goal of local staging is to differentiate patients with tumors confined to the prostate (T₁ and T₂) from patients with tumors that have extracapsular or seminal vesicle invasion (T₃). Patients with organ-confined disease can be treated with surgery or radiation therapy. However, patients with extracapsular extension are generally not surgical candidates and are usually treated with radiation, hormonal, or combination therapy. Accurate preoperative surgical planning aims to decrease the rate of positive surgical margins while maximizing erectile potency rates. Similarly, radiation treatment aims to deliver a specific dose of radiation to a defined volume of tumor while minimizing radiation exposure to surrounding tissues.

Clinical staging, although inexpensive and simple, is not accurate enough for making treatment decisions. Therefore, imaging plays an important role in the staging of prostate cancer. Among the currently available imaging techniques, MRI is the most accurate method for local staging.¹⁰⁰ MRI may be cost-effective, especially for patients who have an intermediate risk of having extracapsular extension and recurrent cancer defined by clinical prognostic factors such as digital rectal examination (stages T₁ and T₂), preoperative PSA (10 to 20 ng/mL), and biopsy Gleason score of 7.¹⁰⁰

Pelvic lymphadenectomy is often part of the initial management of prostate cancer patients because noninvasive imaging may be insufficiently sensitive, particularly for detecting micrometastases. Extensive pelvic lymphadenectomy carries a significant complication rate, however. Sentinel node lymphoscintigraphy is a technique that aims to limit or justify the extent of nodal dissection and duration of surgical procedures—thus limiting complications while preserving the diagnostic sensitivity of surgical staging. A radioactive tracer—usually a micro- or nano-colloid—is injected into tissues in or near the primary tumor; lymphatic channels then drain the colloid to lymph nodes, in which the colloids accumulate. The underyling hypothesis is that the radioactive colloid tracer, following the lymphatic drainage of the tumor, will accumulate in the nodes most likely to contain lymphatic metastases, the so-called “sentinel nodes.” An intraoperative probe is used by the surgeon to detect these radioactive lymph nodes for histopathologic sampling. Limited sentinel node dissection, guided by lymphoscintigraphy and intraoperative probes, has been well validated (i.e., found as accurate as extended dissections) for breast cancer and melanoma. For prostate cancer, tracer-guided selected lymph node dissection appears promising, but a well-designed validation trial involving a large patient population is yet to be conducted.

Testicular Tumors

Testicular tumors are rare, constituting less than 1% of all malignant neoplasms in men, but are the most common malignancy occurring in young men and adolescents.^6,128 They can be classified into germ cell tumors, arising from spermatogenic cells, and non–germ cell tumors, deriving from the sex cords (Sertoli cells) and stroma (Leydig cells). Germ cell tumors constitute 95% of testicular neoplasms and are almost uniformly malignant. However, non–germ cell primary tumors of the testis are malignant in only 10% of cases. The secondary tumors of the testis include tumors such as lymphoma, leukemia, and metastases.¹²⁸

Testicular tumors most commonly present with painless scrotal enlargement or as an incidental testicular nodule. Pain is an uncommon presenting symptom, reported by approximately 10% of patients. In some cases, testicular neoplasia may initially be misdiagnosed as orchitis. Testicular tumors sometimes may undergo regression, necrosis, and scarring (so-called “burned-out” germ cell tumors); therefore, some patients may have normal or small testes at presentation. Some patients, especially those who have an aggressive histologic tumor type, may present with metastases. A small group of patients with hormonally active tumors present with endocrine abnormalities, most commonly gynecomastia.^128,129

Ultrasonography performed with high-frequency transducers is the most commonly used primary imaging modality for the initial workup and evaluation of a scrotal mass, as well as in the staging of a known testicular tumor (Fig. 20-15). It has been shown to be almost 100% sensitive in the detection of scrotal masses. Intratesticular versus extratesticular pathologic conditions can be differentiated with 98% to 100% sensitivity.^128–130 The normal testis demonstrates a homogeneous echotexture. The covering tunica albuginea is generally not seen as a separate structure, except in mediastinum testis, where it can be seen as an echogenic line invaginating into the testis. The epididymis is isoechoic or slightly hyperechoic compared with the testis.

FIGURE 20-15 • Scrotal ultrasound reveals a heterogeneous hypoechoic mass in testis.

Ultrasound is very helpful in the localization of the scrotal mass, whether intratesticular or extratesticular. Further morphologic characterization of the lesion (e.g., as cystic or solid) is also possible. Solid intratesticular masses are frequently malignant. Testicular tumors are usually hypoechoic compared with the surrounding testicular parenchyma. Some tumors can be heterogeneous, containing calcifications or cystic areas. Color Doppler ultrasound may be helpful to identify an isoechoic mass or differentiate a complex cystic lesion from a solid mass.¹²⁸

MRI can be a useful problem-solving tool in the rare cases in which ultrasound is inconclusive.¹³¹ The normal testis demonstrates a homogeneous, intermediate signal intensity on T1-weighted images and high signal intensity on T2-weighted images. The tunica albuginea appears as a thin, low-signal-intensity line surrounding the testis on both T1-weighted and T2-weighted images. The epididymis is isointense or slightly hypointense relative to the testis on T1-weighted images and hypointense on T2-weighted images.^128,132 Testicular tumors often demonstrate homogeneous medium signal intensity similar to normal testis on T1-weighted images and decreased signal intensity within the high signal intensity of normal testis on T2-weighted images. Homogeneous or heterogeneous high-signal-intensity tumors or tumors with signal intensity similar to that of the testis can also be seen.¹³² MRI is very sensitive in depicting the tumors; however, its specificity as to tumor type is low.

Current NCCN guidelines recommend FDG-PET for patients with stage IIB, IIC, and III seminoma who have a residual mass and normal serum markers after primary chemotherapy; in addition, the guidelines state, “for persistent elevated beta-HCG [human chorionic gonadotropin], which is not rising, repeat serial markers, testosterone suppression test, and consider a PET scan.”¹³³ Gonadal germ cell tumors comprise several histologic subtypes,¹³⁴ possibly with differing FDG-avidities.^135–138 Overall, FDG-PET appears sensitive in detecting seminoma¹³⁹ and nonseminomatous germ cell carcinoma (NSGCT).¹⁴⁰ A recent statement by the European Germ Cell Cancer Consensus (EGCCC) group¹⁴¹ states that FDG-PET does not improve staging of germ cell cancers compared with CT alone; to support this, the authors cite articles that show FDG-PET does improve staging compared with CT alone^136,142,143 and is more sensitive than CT.^136,143,144 Similarly, the EGCCC group makes the potentially misleading statement that PET scans are not recommended as part of routine initial staging procedures; the citations following this statement (see original EGCCC article¹⁴¹) include no articles that recommend against the use of PET in initial staging. Some articles cited by the EGCCC used PET methodology no longer accepted as state-of-the art (e.g., filtered backprojection data reconstruction^145,146). A rationale for early detection of tumor relapse in NSGCT is to “catch” recurrent disease in a stage with a more favorable prognosis.¹⁴⁷ In stage I NSGCT patients with negative CT and negative tumor markers who were postorchiectomy, FDG-PET/CT identified metastatic disease in approximately 15% to 21% of cases.^143,144 Yet in patients with stage I NSGCT who have a high risk for recurrence, a negative FDG-PET scan does not modify or substratify the risk of relapse.¹⁴⁴ As suggested by the NCCN guidelines, in the postchemotherapy setting FDG-PET appears able to distinguish viable germ cell tumor from inflammatory and necrotic debris (which are often somewhat FDG-avid).^{137,139,140,148,149}

FDG-PET accuracy appears lower in patients with metastatic NSGCT and postchemotherapy masses.¹⁴⁰ FDG-PET is incapable of distinguishing mature teratoma from inflammatory residuum, using standard semiquantitative (standard uptake value) PET measurements^150,151; postchemotherapy masses containing mature teratoma are frequently false-negative.^135,140,150 In general, with the conventional approach to FDG-PET, it appears that a positive FDG-PET scan is highly reliable in diagnosing residual tumor, whereas a negative FDG-PET is less reliable.¹⁵²

Sugawara and colleagues found that more sophisticated quantitative analysis of FDG tumor pharmacokinetics distinguishes mature teratoma from inflammatory residuum¹⁵¹; yet the findings remain to be confirmed. Tissue inflammation related to therapy and tumor death is an FDG-avid process. In a small group of patients, Haba and colleagues¹⁵³ attempted to circumvent the problem of FDG-avid inflammation obscuring tumor evaluation: in patients with stage 2 NSGCT, the investigators acquired FDG-PET scans 72 to 96 hours after initiation of chemotherapy, presumably aiming for a time-point after cancer cytotoxicity and cell death but before significant tissue infiltration by the FDG-avid inflammatory cells that clear tumor debris. Four of the six patients’ scans demonstrated complete resolution of tumor FDG-uptake, and one patient’s scan demonstrated a 50% reduction; these five patients went on to demonstrate a complete response by conventional criteria. Validation in a larger patient population is needed.

Penile Carcinoma

Carcinoma of the penis is a rare disease. Most penile cancers are squamous cell carcinomas. Penile carcinoma’s causal factors are unclear, but it is seldom seen in circumcised men and it is related to the presence of foreskin, irritant effects of smegma, and poor hygiene. Treatment depends on the degree of involvement and ranges from partial penile amputation to total penectomy.¹⁵⁴ MRI is a useful tool to evaluate penile tumors that appear iso- to hypointense on T1-weighted images and hypointense on T2-weighted images compared with the surrounding corporeal body. MRI is especially useful when the tumor extends into the radix penis, which is difficult to evaluate clinically. Depiction of tumor extension into the bulb of the penis or across the urogenital diaphragm is important in determining treatment options, which include radiation or surgery.¹⁵⁴

Not much clinical data has been reported regarding FDG-PET for penile cancer, yet it appears that untreated, primary penile cancers are frequently FDG-avid and that FDG-PET/CT is highly accurate for lymph node staging.¹⁵⁵

Bladder Cancer

Bladder carcinoma is the most common malignancy involving the urethlial tract. It occurs more commonly in males than in females.⁶ Most bladder cancers are of epithelial origin and up to 90% are transitional cell carcinomas. Squamous cell carcinomas (5% to 10%) and adenocarcinomas (2%) are the other types of epithelial bladder cancers. Squamous cell carcinomas are associated with chronic inflammatory diseases of the urinary bladder.¹⁵⁶ The prognosis of urinary bladder cancer depends on histologic type, pattern of tumor growth, and anatomic extent of disease at the time of diagnosis. Most patients present with superficial disease, for which transurethral resection and multiple cystoscopic biopsies are sufficient for diagnosis, staging, and definitive therapy. Staging with cross-sectional imaging is necessary when deeply invasive tumors are detected (Fig. 20-16 and Fig. 20-17). Evaluation of bladder cancer consists of determining the site, number and growth pattern of the tumors, depth of muscular and perivesical invasion, and lymph node status.

FIGURE 20-16 • Contrast-enhanced axial computed tomography image reveals bladder cancer causing thickening of the left posterolateral wall.

FIGURE 20-17 • Axial T2-weighted magnetic resonance image depicts bladder cancer on left posterolateral wall. Note that the normal low signal intensity of the bladder wall is preserved, excluding tumor invasion.

On cross-sectional imaging, it is difficult to differentiate tumors that are confined to the mucosa and submucosa (stage T₁) from tumors superficially invading the muscle wall (stage T₂). Deep muscle invasion (stage T_3a) is diagnosed when there is interruption of the urinary bladder wall. Perivesical fat invasion (stage T_3b) can be diagnosed when there is infiltration in the perivesical fat. However, inflammatory changes can mimic tumor invasion into perivesical fat and may cause over-staging. Direct invasion of adjacent organs (stage T₄) can also be easily seen on CT or MRI. Detection of lymph node metastasis depends on demonstration of enlarged regional lymph nodes. However, metastatic lymph nodes cannot be differentiated from hyperplastic ones.

Different numbers of staging accuracies for bladder carcinoma were reported with CT (40% to 92%) and with MRI (50% to 96%). Dynamic contrast-enhanced MRI provides increased tumor detection and staging accuracy for superficial lesions. Limitations of both MRI and CT include their inabilities to depict small tumor growth into the muscle layer of the bladder wall or differentiate between tumor and inflammatory changes.^157–163

Current NCCN guidelines do not mention a role for FDG-PET in the management of bladder cancer.¹⁶⁴ Currently, no original research articles on bladder cancer and state-of-the-art FDG-PET/CT are available; a trial of modern FDG-PET/CT in bladder cancer ought to be done. Evaluation of primary tumors in bladder cancer by FDG-PET is generally precluded by the presence of excreted FDG in the urine; the intensity of radioactivity in the urine can obscure evaluation of adjacent nodal regions.¹⁶⁵ This problem can be addressed by the use of indwelling bladder catheters and irrigation,¹⁶⁵ diuresis,^166,167 delayed imaging,¹⁶⁶ and clever use of intravenous radiographic contrast and patient positioning.¹⁶⁸ For lymph node staging, FDG-PET has a sensitivity of 67% to 76% and specificity of 86% to 97%, although some of the published studies have used outdated PET methodology.^{126,169–171}

Rectal Cancer

Colorectal carcinoma is the most common cancer of the gastrointestinal tract, and the second most common cause of cancer-related death in the United States.⁶ The most common histologic type of sigmoid and rectal cancer is adenocarcinoma. The standard surgical management for rectal carcinoma was once abdominoperineal resection and formation of permanent colostomy. However, new surgical techniques, such as low-anterior resection with anal sphincter sparing and local excision, have become more popular for well-differentiated, early-stage tumors. Preoperative staging of rectal cancer determines the choice of surgical approach and the need for adjuvant radio- or chemotherapy. In addition, the degree of tumor invasion through the layers of the rectal wall and the patient’s nodal status are very important features that determine the risk of local tumor recurrence.^173,174 For these reasons, correct preoperative local staging of rectal cancer is essential.

Transrectal ultrasound and MRI are commonly used for assessing tumor penetration through the rectal wall layers and nodal status. Unlike CT, both transrectal ultrasound and MRI can distinguish the layers of the rectal wall and allow depiction of tumor growth through these layers. According to the TNM staging system, a stage T₁ tumor is limited to the mucosa and submucosa and does not extend to the muscularis propria. Stage T₂ refers to tumors that partially involve the muscularis propria. A stage T₃ tumor totally disrupts the muscular layer and extends into the perirectal fat. When invasion is present in adjacent organs or structures, the tumor is identified as stage T₄ (Fig. 20-18). Rectal cancer is an FDG-avid disease,¹⁷⁵ although the degree of FDG-avidity can be variable, and not clearly associated with the size of the tumor¹⁷⁶ or the proliferative rate of its cancer cells.¹⁷⁷ Two major factors, in determining the FDG-avidity of a rectal tumor are its density of cancer cells and the hexokinase activity of those cells.¹⁷⁸ Although nonmucinous adenocarcinoma is FDG-avid on PET, mucinous cancer can lack distinct FDG uptake, likely because of the relative hypocellularity of such tumors.¹⁷⁹ Premalignant polyps can be FDG-avid as well, and other FDG-avid diseases of the rectum can mimic a tumor on FDG-PET, including abscesses and focal rectal inflammation caused by prior polypectomy.¹⁷⁶

FIGURE 20-18 • Axial (A) and sagittal (B) T2-weighted magnetic resonance images reveal rectal cancer causing thickening in the rectal wall and invading the vagina.

For the local staging of rectal cancer, the reported accuracy of transrectal ultrasound ranges between 70% and 81%, and that for MRI between 71% and 91%. Most under-staging errors are caused by the inability of both methods to detect microscopic or minimal invasion. Inflammatory changes in perirectal fat mimicking tumor invasion are reported to be the most frequent cause of over-staging.^180–187

The 2008 NCCN guidelines state that FDG-PET is not routinely recommended in the workup of rectal cancer appropriate for resection.¹⁸⁸ Yet the addition of FDG-PET to conventional staging changes clinical management in approximately 17% to 27% of cases^{101,175,189–193} (Fig. 20-19). FDG-PET findings may help clinicians avoid futile surgery or adjust radiotherapy fields.¹⁹⁰ Compared with conventional CT-guided planning of IMRT, FDG-PET changes the radiation treatment plan in approximately 26% of cases.^189,190 A small study suggested that PET-CT reduces the interobserver variability of target-volume delineation in rectal cancer, compared with CT-guided planning.¹⁹⁴ One controversial issue in the use of FDG-PET/CT for radiotherapy planning is the proper method of adjusting the PET image intensity windows so that target volume is accurately delineated.^195–197 Target volumes are an anatomic issue and FDG-PET is not an anatomic imaging modality in its essence. FDG-PET may provide a coarse scintigraphic estimate of tumor volume; PET image intensity windows can be adjusted so that the region (i.e., voxels) of FDG uptake closely matches the true anatomic volume of tumor. The question remains how to do this accurately without foreknowledge of true tumor volume. The group at St-Luc University Hospital, in Brussels, Belgium, provided one attractive, semiautomated computer algorithm intended for this purpose,¹⁹⁸ which they then validated against surgical specimens in pharyngolaryngeal squamous cell carcinoma.¹⁹⁹ Of note, in this study, FDG-PET was found to be the superior modality for delineating target volume, compared with CT and MRI.

FIGURE 20-19 • Pretreatment staging fluorodeoxyglucose–positron emission tomography/computed tomography (FDG-PET/CT) of 59-year-old female with colorectal cancer. At left: lateral view of a maximum-intensity-projection PET image, demonstrating FDG uptake in two synchronous malignant rectosigmoid neoplasms (arrowheads; grey arrow: physiologic bladder activity), with no evidence of metastatic disease. Axial CT (middle image) showed enlarged pararectal lymph nodes (e.g., 1.2-cm diameter node at white arrow) that were not FDG-avid (as shown in fusion image, at right). Surgical staging confirmed two rectosigmoid malignant neoplasms and absence of locoregional metastases.

FDG-PET/CT is most accurate in the pretherapy setting,^178,200 being extremely sensitive in the detection of liver and lung metastases,²⁰⁰ the two viscera most commonly involved by metastatic disease at initial presentation.²⁰¹ In general, FDG-PET/CT has long been recognized as superior to conventional CT for detecting hepatic metastases,^176,202,203 although its sensitivity decreases in characterizing subcentimeter hepatic lesions.^204,205 FDG-PET is recommended by the NCCN for the workup of patients with metachronous resectable metastases.¹⁸⁸ FDG-PET will detect extrahepatic disease in 18% to 32% of patients planning to undergo hepatic resection based on conventional imaging; in 20% to 40% of these patients with extrahepatic disease, it will lead to a change in management.^204–207 A recent study from Washington University²⁰⁸ found that patients with hepatic metastases who underwent FDG-PET for preoperative staging had a much higher 5-year survival rate than did historic controls. Yet structural imaging will continue to have an important role, as it provides indispensable anatomic information for surgical planning.

Dedicated chest CT is likely more sensitive than FDG-PET/CT for the detection of small lung metastases.²⁰⁹ The companion CT of a PET/CT study is optimized for PET imaging—not for imaging pulmonary anatomy—making it more difficult to detect subcentimeter pulmonary nodules. Such tiny lesions are often nonspecific and require surveillance with optimized chest CT.

For patients with stage II or III rectal cancer, treatment with preoperative combined-modality therapy (CMT) followed by total mesorectal excision is a common approach²¹⁰; hypothetically, it would be advantageous to know how well tumors respond to preoperative CMT early in the course of treatment, so that ineffective treatments could be changed.²¹¹ Yet ultrasound, MRI, CT, and digital rectal examination all do a suboptimal job of identifying tumor response to preoperative CMT.^212–215 A number of studies have suggested that FDG-PET is the best noninvasive modality for identifying the presence and extent of rectal tumor response, correlating with histopathology^{177,211,213,216–219} and prognosis.^192,220 Most studies on these methods examined rectal tumor FDG uptake before and after the completion of preoperative therapy, although one exception studied therapy-induced changes during treatment (day 12) and found one of the highest accuracy levels reported.²¹⁸ A recent Danish study found that FDG was not accurate in identifying the presence and extent of rectal tumor response to therapy²²¹; however, in the Danish study, FDG-PET was performed only at a single time-point (post-treatment) and therapy-induced changes in FDG uptake were therefore not examined. Therapy-induced reductions in rectal tumor FDG uptake appear to mainly reflect reductions in the tumor cancer cell population¹⁷⁷ and cellular hexokinase activity,¹⁷⁸ rather than changes in cellular proliferation.¹⁷⁷ Post-therapy tumor inflammation is a recognized phenomenon^222,223 and it is FDG-avid, potentially obscuring the signal loss associated with the death of FDG-avid cancer cells, by FDG-PET. Tumor inflammation is likely a good sign in terms of patient prognosis: the Memorial Sloan-Kettering Cancer Center group reported that a paucity of inflammatory or ulcerative changes in tumors after preoperative chemoradiotherapy is an independent predictor of a worse outcome.²²⁴

The American Society of Clinical Oncology,²²⁵ the European Society for Medical Oncology,²²⁶ and the NCCN¹⁸⁸ do not currently recommend PET or MRI for routine tumor surveillance in the management of rectal cancer. Yet the NCCN guidelines do recommend that physicians consider obtaining an FDG-PET scan in the setting of a rising serum carcinoembryonic antigen (CEA) level. Serum CEA levels play a key role in surveillance, as rises in CEA levels precede detectable abnormalities on conventional imaging, typically by 5 to 8 months.^227,228 Yet several studies have found that FDG-PET/CT is highly accurate for detection of recurrent disease and alters patient management in a significant percentage of cases (26%-61%).^{179,229–232} In studies of colorectal cancer patients, FDG-PET has repeatedly proven its usefulness in the setting of rising CEA when conventional imaging studies were normal^233–236 or equivocal.²²⁹ Moreover, prospective comparisons^236,237 of FDG-PET and spiral CT in colorectal cancer patients with suspected recurrence found FDG-PET superior for both staging local recurrence and detecting hepatic metastases; in this setting, the use of FDG-PET changed clinical management in 38% of cases,²³⁷ often preventing futile surgery, to the benefit of the patient as well as the healthcare system.²³⁶ Retrospective studies yielded similar findings.^238–241

If conventional imaging documents metachronous metastases that appear resectable, NCCN guidelines recommend evaluation by FDG-PET.¹⁸⁸ A well-designed study by the National Institutes of Health (NIH),²³⁴ albeit with a small study population, addressed the following issue: 60% to 90% of patients with CEA elevations and normal conventional structural imaging findings have recurrent disease,^242,243 yet at laparotomy, only half of these recurrences will prove resectable (resectability being associated with a more favorable prognosis)²⁴⁴; hence, the main role of imaging is not for detecting recurrence (CEA levels being more sensitive than imaging in this regard), but for defining recurrences as resectable or nonresectable. The NIH group studied preoperative FDG-PET in two groups of patients staged by conventional imaging: those with no imaging evidence of disease and those with recurrence that appeared resectable on conventional imaging. Extra-abdominal disease, visualized by PET, was confirmed by biopsy; patients without extra-abdominal disease on PET went to laparotomy. Two surgeons, one blinded and one unblinded to PET information, inspected the abdominal cavity sequentially. Of the 15 patients thought to have resectable disease based on conventional imaging, 6 were found to be unresectable at laparotomy; 11 patients with occult disease on imaging were found to have intraabdominal recurrences at laparotomy, of which 4 were unresectable. Of these 10 cases of unresectable disease in patients who went to laparotomy, preoperative FDG-PET had correctly identified 90% as being unresectable. Another study reported similar accuracy for FDG-PET in predicting resectability.²³⁶ Of the 16 cases of resectable disease in the NIH study, preoperative FDG-PET incorrectly identified 19% as unresectable. FDG-PET, when read by imaging specialists blinded to conventional imaging findings, located tumor in all patients with known recurrence; in patients with intra-abdominal recurrences that were invisible to conventional imaging, FDG-PET detected 89% of tumors found at laparotomy by the two surgeons. Moreover, in the 11 patients with intra-abdominal recurrences that were invisible on conventional imaging, the blinded surgeon missed two recurrences that were detected with FDG-PET and were found by the surgeon who operated with the benefit of knowing the PET results. Of the two patients without recurrent colorectal cancer, at laparotomy and confirmed by 2 to 4 years of follow-up, FDG-PET correctly excluded disease in one, while incorrectly identifying recurrent disease in the other. Thus, preoperative FDG-PET provides important guidance to surgeons in locating recurrences at laparotomy; identification of extra-abdominal disease may allow avoidance of futile surgery, but the fact that FDG-PET incorrectly identified 19% of resectable abdominal recurrences as unresectable raises doubt as to its reliability for deciding to forgo surgery, as removal of resectable disease improves survival. Moore and colleagues reported similar sensitivity (84%) for FDG-PET using modern techniques in the detection of pelvic recurrence of rectal cancer, with specificity of 88%.²⁴⁵ It must be noted that the NIH study and the study by Moore and colleagues used FDG-PET, not FDG-PET/CT. Cohade and colleagues²⁴⁶ found that PET/CT improved diagnostic accuracy, compared with FDG-PET alone, in a population of colorectal cancer patients. In rectal cancer patients, Even-Sapir and colleagues²⁴⁷ found PET/CT superior to PET alone, as physiologic uptake in displaced viscera sometimes fooled readers of PET images without CT; the authors reported that FDG-PET/CT had 98% sensitivity and 96% sensitivity in the detection of pelvic recurrence on a per-lesion basis.

Even-Sapir and colleagues found FDG-PET/CT extremely accurate in distinguishing viable tumor from postoperative scarring in the presacral region, with 100% sensitivity and 96% specificity. Fukunaga and colleagues²⁴⁸ fused independent PET and CT images using software, and found that software-fused FDG-PET/CT had diagnostic accuracy of 93%, which was superior to that of CT alone; in particular, FDG-PET distinguished viable tumor from postoperative scarring. The complementary information altered management in 26% of cases. Yet software-fused PET/CT is not the same as true PET/CT acquired with hybrid PET/CT scanners^249,250; the University of California, Los Angeles, group demonstrated that recurrence of colorectal cancer is detected by “in-line” hardware-fusion PET/CT with significantly greater accuracy than it is with software-fused PET/CT.²⁴⁹

Kinkel and colleagues²⁰³ reported a meta-analysis comparing CT, MRI, ultrasound, and FDG-PET for the detection of hepatic metastases as sites of relapsed gastrointestinal cancers. FDG-PET had a sensitivity of 90% (95% confidence interval [CI] 80%-97%) that was significantly superior to the sensitivities of all other modalities. Whiteford and colleagues reported PET-CT to be superior to conventional CT in detecting hepatic and extrahepatic metastases in recurrent rectal cancer.¹⁷⁹