Anorectal Anatomy

The rectum extends from the upper end of the anal canal to the rectosigmoid junction and is approximately 12 cm in length. It is subdivided into proximal, middle, and distal thirds, depending on the distance of the most distal aspect of the tumor from the anal verge. The surgical anal canal extends from the anorectal junction until the anal verge and measures between 2.5 and 4 cm in length. The anatomic anal canal corresponds to the distal two-thirds of the surgical anal canal and is separated from the proximal one-third by the dentate line. Above the dentate line, the anal canal is lined with columnar epithelium, whereas it is lined with squamous epithelium distal to the dentate line. The anal transitional zone corresponds to an approximately 10-mm area between the columnar and squamous epithelial zones where the mucosa is of variable histology.

The rectal wall is composed of mucosa, submucosa, and muscularis propria. The mucosa and submucosa complex appears as a three-layered wall structure on EUS. The mucosa is composed of two wall layers: an inner hyperechoic layer (the interface between the mucosa and the ultrasound probe) and an outer hypoechoic wall layer. This is accompanied by the third wall layer, which is hyperechoic, representing the submucosa. The muscularis propria of the rectum, or fourth wall layer, is composed of an outer longitudinal and inner circular smooth muscle layer. The inner circular smooth muscle becomes thickened distally and continues as the internal anal sphincter. The outer longitudinal muscle fuses with fibers from the levator ani. The outermost layer of the sphincter complex is formed by striated muscles: the levator ani and puborectalis muscles superiorly and the inferior part of the external anal sphincter inferiorly.

The rectum is surrounded by mesorectal fat containing lymph nodes, superior hemorrhoidal vessels, and fibrous tissue collectively known as the mesorectum. The mesorectum is continuous with the fat of the sigmoid mesocolon superiorly and is usually thicker along the posterior rectum in its intraperitoneal portion; on occasion it is absent anteriorly. It is bound circumferentially by the mesorectal fascia. This fascia extends inferiorly and coalesces with the Denonvilliers fascia in men, and anterior to it are the seminal vesicles and the prostate gland. Conversely, in women the anterior mesorectal fascia coalesces with rectovaginal fascia, anterior to which is the vagina. The mesorectal fascia forms an important barrier to the radial spread of upper and middle third rectal tumors and forms the plane of dissection used in total mesorectal excision (TME).

Nodal drainage of the rectum occurs initially to the perirectal lymph nodes within the mesorectum. The majority of such nodes follow the rectal blood supply and are located superiorly and posteriorly. Common nodal spread is along the superior rectal artery into the apical mesorectum and the inferior mesenteric artery into the sigmoid mesocolon. The middle rectal artery arises from the internal iliac artery directly, and the inferior rectal artery arises from the internal pudendal artery, which is a branch of the anterior division of the internal iliac artery. The inferior and middle rectal arteries anastomose at the anorectal junction, and although uncommon, distal rectal cancers can spread to the nodes along the internal pudendal and internal iliac arteries.

Rectal Cancer Tumor-Node-Metastasis Staging

The tumor-node-metastasis (TNM) system advocated by the American Joint Committee on Cancer (AJCC) and the International Union Against Cancer (UICC) have become the worldwide standard for staging colorectal cancer. The TNM system classifies the extent of the tumor (T stage) by the depth of tumor invasion into and through the rectal wall. Nodal substations classified as regional lymph nodes for rectal cancer are perirectal, sigmoid mesenteric, inferior mesenteric, lateral sacral, presacral, sacral promontory, internal pudendal, internal iliac, superior rectal, middle rectal, and inferior rectal. The involvement of lymph nodes outside these groups, such as in the external or common iliac substations, is considered to be distant metastases (M stage; Table 18.1 ).

Other tumor characteristics that are important to consider for imaging purposes include the proximal and distal tumor margins, the extent of tumor annularity, the presence of ulceration, anal sphincter complex invasion, and the relationship of the distal tumor margin to the middle valve of Houston. The valve of Houston is thought to be a surrogate marker for the anterior peritoneal reflection, and the location of a tumor proximal or distal to the anterior peritoneal reflection has important surgical planning implications.

T-Staging Considerations

Rectal cancer usually appears as a hypoechoic lesion that disrupts the normal five-layer sonographic structure of the rectal wall. It is important to document where the distal border of the tumor is in relation to the seminal vesicles in males and the cervix in females, in order to clarify the lesion location in relation to the anterior peritoneal reflection. This is then compared with the endoscopic estimate of the distal tumor border. In published studies, the accuracy of EUS T-staging ranges from 80% to 95%, compared with 65% to 75% for CT and 75% to 85% for MRI ( Fig. 18.1 ). With respect to T stage, one particular problem is the overstaging of T2 tumors due to the difficulty in differentiating peritumoral inflammation secondary to a desmoplastic reaction from tumor fibrosis ( Fig. 18.2 ).

A superficial primary rectal cancer (T1N0) on the distal valve of Houston in an 84-year-old male, managed conservatively by a snare resection.

An ulcerated friable distal primary rectal cancer (T2N0) in a 62-year-old female who proceeded directly to an abdominoperineal resection. The tumor invaded the muscularis propria (hypoechoic fourth endoscopic ultrasonography layer) but did not penetrate through it (T3) or extend beyond the five echo layers into the surrounding perirectal tissue.

A T3 tumor must extend through the entire thickness of the muscularis propria into the perirectal fat, obliterating the sharp fat–muscle interface with features of pseudopodia ( Video 18.1 ). It is thought that all T3 rectal tumors are not equal, with minimally invasive disease carrying a more favorable prognosis. Therefore by discriminating minimally invasive from advanced T3 disease (invasion ≤2 or >3 mm beyond the muscularis propria), preoperative EUS may provide important prognostic information. However, the challenge is that overstaging is noted to be more common in minimally invasive T3 (50%) when compared with advanced T3 disease. A maximum tumor thickness measured in a T3 cancer is also an independent prognostic factor for local and overall recurrence. A maximum tumor thickness cutoff measurement ≥19 mm has been proposed to be useful when classifying patients preoperatively and to select patients for primary surgery or neoadjuvant therapy.

Video 18.1

Staging of Rectal Cancer as T3NI by Endoscopic Ultrasonography

Conversely, understaging may be caused by a failure to detect microscopic cancer infiltration, owing to the limits of EUS resolution. Resolution is improved by increasing ultrasound frequency but at the expense of a reduction in the depth of penetration, such that it may be impossible to visualize the leading edge of a tumor. This may limit the detection of invasion of adjacent organs. Important variables that influence the accuracy of tumor staging include operator experience and the location of the tumor within the rectum, with reduced accuracy for more distal tumors.

A meta-analysis of 42 studies ( n = 5039 patients, from 1980 to 2008) that reviewed the EUS accuracy when differentiating T stages suggested that EUS sensitivity is greatest for advanced disease compared with early disease ( Table 18.2 ).

N-Staging Considerations

Conventional EUS nodal echo features that accurately predict nodal metastasis have been identified in patients with esophageal cancer. These ultrasound features include lymph node short axis size, echogenicity, shape, and border. Features proposed to correlate with malignancy include an enlarged node (≥1 cm in short axis), hypoechoic appearance, round shape, and smooth border ( Fig. 18.3 ; Table 18.3 ). However, these conventional EUS nodal criteria have proven inaccurate for staging many nonesophageal cancers. No single criterion is predictive of malignancy in patients with lung, esophageal, and pancreatic cancer. If all four abnormal morphologic features are present, the accuracy for malignant invasion is 80%. However, all four features of malignant involvement are present in only 25% of malignant lymph nodes ( Table 18.4 ).

A T3N1 lesion in a 54-year-old male who proceeded to neoadjuvant therapy followed by surgery (xxxx = tumor breaching through the fourth muscularis layer and making the lesion a T3 lesion). The highlighted node is perilesional and therefore not amenable to fine-needle aspiration. It has a hypoechoic appearance and short axis greater than 5 mm but is oval in shape with an irregular border.

Although EUS FNA is the most accurate modality for locoregional staging of cancer, the N-staging accuracy is only 70% to 75% and was recently reported to be as low as 42%. It was previously assumed that EUS was incapable of detecting benign perirectal lymph nodes. Therefore in patients with rectal cancer, the visualization of lymph nodes was considered to be an accurate surrogate marker of nodal metastasis, thereby obviating the need for FNA. A meta-analysis (35 studies, n = 2732 patients, from 1966 to 2008) that reviewed the literature regarding N stage EUS accuracy suggested that the sensitivity and specificity of EUS is moderate and that further refinements in diagnostic criteria are needed to improve the diagnostic accuracy. It is important to note that all of these studies were non-FNA, primarily radial EUS examinations.

Prior transrectal ultrasound studies identified a nodal size of ≥7 mm as an optimal size cutoff for predicting nodal metastases in rectal cancer, with an accuracy of 83% when compared with surgical pathology. A dedicated FNA study based on a perception that metastatic locoregional nodes are only minimally different in morphologic appearance when compared with benign nodes noted that the number of conventional malignant echo features per lymph node did not accurately differentiate benign from malignant nodes, unless all four features were present. The accuracy of conventional criteria to include short axis ≥10 mm, hypoechoic appearance, round shape, and smooth border for detecting malignant lymphadenopathy was 61%, 65%, 51%, and 51%, respectively. A lymph node short-axis length ≥5 mm or hypoechoic appearance was the only conventional features predictive of malignant infiltration. An optimum short- and long-axis length of 6 and 9 mm yielded the best power distinction for malignancy. Using surgical histopathology specimens, Knight and colleagues assessed the performance characteristics for overall sensitivity, specificity, and positive and negative predictive values of FNA in the setting of primary or metastatic colorectal carcinoma, reflecting values of 89%, 79%, 89%, and 79%, respectively.

The preoperative FNA identification of extramesenteric lymph node metastases upstages 7% of primary rectal cancers undergoing an EUS evaluation. For example, external iliac lymph node infiltration is outside the standard operative field for TME. This location, if recognized at EUS, may impact medical and surgical planning by altering the standard radiation fields or may alter surgical planning to extend the TME resection to include an extensive lymph node dissection. Significant clinical, endoscopic, and sonographic features associated with such metastases include serum carcinoembryonic antigen (CEA) level, tumor length ≥4 cm, tumor annularity ≥50%, sessile morphology, and lymph node size.

The recent findings indicate that FNA should be used when verifying nodal status and when making critical decisions regarding the use of neoadjuvant therapy rather than relying on nodal appearance alone. Failure to use FNA risks stage-inappropriate therapy and, in turn, inappropriate patient outcomes. A note of caution is that luminal fluid cytology may be positive for malignancy in 48% of luminal cancers, including rectal cancer, but is not affected by performing FNA. This translocated cell contamination, along with endosonographer technique and cytologic misinterpretation, are risk factors for false-positive EUS FNA cytology.

EUS FNA of solid lesions in the lower GI tract is considered to be a low-risk procedure for infectious complications and does not warrant prophylactic administration of antibiotics for the prevention of bacterial endocarditis. Until adverse event data become available, perirectal cystic structures should not be sampled, as abscess formation requiring percutaneous drainage has occurred despite the administration of prophylactic antibiotics.