Материал: Advanced Imaging of the Abdomen - Jovitas Skucas

be obtained from a tumor showing obvious malignant imaging characteristics is a matter of opinion and established local practice. Most biopsies of primary colon carcinomas are obtained via colonoscopy (or sigmoidoscopy for rectosigmoid lesions), although percutaneous biopsy using imaging guidance is feasible for larger tumors.

Detection

Barium Enema: The relative roles of barium enema, flexible sigmoidoscopy, and colonoscopy are not established. A number of studies have shown a superiority of colonoscopy over barium enema. The problem with most of these studies is that colonoscopy is used as a gold standard and the studies are performed by gastroenterologists, but barium enemas were performed by general radiologists and the results are often biased against barium enema. A barium enema, however, does detect most pedunculated, sessile and infiltrating colon (Fig. 5.23) and rectal (Fig. 5.24) carcinomas.

In a retrospective multihospital Indiana study, the sensitivity of colonoscopy for detecting colorectal cancer (95%) was greater than with a barium enema (83%) (126); the sensitivity of a double-contrast barium enema (85%) was no different from that of a single-contrast study (82%). Barium enema performed no better in the right than the left colon. Cancers

ADVANCED IMAGING OF THE ABDOMEN

detected by colonoscopy were more likely to be Dukes’ class A (25%) than cancers detected by barium enema (10%).

A retrospective colon cancer study from a well-defined geographic region in Norway found that a barium enema correctly detected a cancer in 91% of 386 tumors, a cancer or major precancerous lesion was overlooked in 7%, and the examination was not possible in 2% (127); colonoscopy, on the other hand, correctly detected cancer in 80% of 215 tumors, cancer or a major precancerous lesion was overlooked in 6%, and colonoscopy was technically incomplete in 14%.

Endoscopy (Conventional Colonoscopy): In an Indiana study, colonoscopy performed by gastroenterologists was more sensitive (97%) for cancer detection than those done by nongastroenterologists (87%) (126).

How accurate is colonoscopy in localizing a colorectal cancer? One study of 77 cancers revealed significant errors in tumor localization in 8% (a significant error was defined as a change from the preoperative planned resection to an alternative resection) (128). A retrospective study of colorectal cancers not detected by colonoscopy performed within 3 years of diagnosis suggested that 57% of the cancers were “missed,” and 43% were believed not to have been reached, although some right colon cancers recorded as missed may have been not reached (129); the authors suggested that cecal

intubation should be verified by specific landmarks in all instances, and failure to reach the cecum should be followed by a prompt barium enema or CT colonography.

Computed Tomography: Computed tomography of some large, fungating ascending colon carcinomas infiltrating to pericolic fat identifies segmental distal colonic wall thickening. The histopathology of resected specimens reveals submucosal and subserosal edema, chronic inflammation and fibrosis, or both (130).

Not all focal colorectal tumors detected by CT or MR are neoplastic. An adjacent abscess can readily mimic a necrotic cancer and vice versa (Fig. 5.25). Endometriosis is another example.

Intravenous contrast enhancement aids polyp detection (131); both benign and malignant polyps enhance with contrast, while residual content does not. Enhancement significantly improves visualization of 6–9 mm polyps (75% postcontrast versus 58% precontrast) (132). Contrast also aids detection of local tumor extension and any lymphadenopathy. Most published performance data on CT colonoscopy are summarized in a 2003 book on this topic (133).

Studies suggest that CT colonography is competitive with conventional colonoscopy in detecting both benign and malignant polyps >1cm. In patients with colonic tumors (confirmed at endoscopy or surgery), axial and multiplanar CT detected all malignancies (134); all missed benign tumors were <8mm in diameter. Computed tomographic colonoscopy

performed the same day as conventional colonoscopy achieved a 58% sensitivity and 52% specificity in identifying polyps, with sensitivity for polyps ≥1cm being 86% (135).

After bowel preparation and colon air insufflation, 300 patients underwent CT scanning in supine and prone positions using 3-mm collimation and single breath hold (136); transverse CT images, sagittal and coronal reformations, and 3D endoluminal images completed the CT colonography. Using conventional colonoscopy results as a gold standard, this study achieved a sensitivity of 90% for detecting polyps 10mm or larger, 80% for polyps 5.0 to 9.9mm, and 59% for polyps <5mm; of note is that CT colonography detected all carcinomas. Intravenous contrast provides colonic wall and tumor enhancement. Enhancement significantly improved visualization of 6- to 9-mm polyps (75% postcontrast versus 58% precontrast) (9). One potential pitfall for CT colonography is the occasional carpet-like (137) or flat cancer. Most studies suggest that multiplanar 3-D endoluminal images achieve better sensitivity and specificity than 2-D images; nevertheless, in any one patient a combination of images is often necessary for full evaluation.

Computed tomographic colonoscopy is an alternative to barium enema and conventional colonoscopy, especially in frail, elderly patients. Detection of small polyps in these patients is not as relevant as in younger patients.

Ultrasonography: Endorectal US detects rectal tumors. Attempts have been made to

detect a malignancy arising within a rectal villous adenoma, but results have been disappointing. In general, rectal villous adenomas are resected regardless of imaging or biopsy findings.

Uncommon Type/Presentation

Flat (Depressed) Adenomas and Carcinomas: Socalled flat (also called depressed and superficial

depressed) colon adenomas and carcinomas do not have a predominant intraluminal growth pattern; rather, they show a tendency toward early submucosal invasion and early metastasis. Nevertheless, they grow slowly. A retrospective collection of nine flat colon carcinomas found an initial mean 12-mm diameter, and it took these cancers an average of 32 months to double in size; a comparable sample of polypoid carcinomas doubled in size, on average, in 9 months

COLON AND RECTUM

(138); with time, flat cancers continued with a nonpolypoid growth pattern. Cell kinetics and molecular alterations in flat tubulovillous tumors suggest that they represent a distinct entity differing from their polypoid counterpart (139). Some evidence suggests that flat colonic tumors evolve from colonic mucosa overlying lymphoid nodules.

These cancers are less common than those evolving via the adenoma-carcinoma sequence. A higher proportion develop in a setting of inflammatory bowel disease and radiation proctocolitis. Some are rather aggressive and when still small have already metastasized to the liver, but these are rare exceptions.

Comparing flat adenomas and adenocarcinomas evaluated in Stockholm and Tokyo by the same pathologist, the Japanese lesions were more advanced with regard to dysplasia and were more aggressive (140), suggesting the presence of different geographic manifestations. Complicating the picture, in the United States a number of small, flat umbilicated tumors are hyperplastic polyps rather than neoplasms.

Although many radiologists believe that even a technically excellent double-contrast barium enema does not detect most flat colonic neoplasms, a Japanese study suggests otherwise (141); among 97 early flat and depressed colorectal cancers, a double-contrast barium enema detected converging folds and semilunar deformity more often in cancers with moderate- to-massive submucosal extension than in those confined to mucosa or with only focal submucosal extension. Also, deep depressions, an irregular surface in these depressions, and tumors >20mm were predictive of submucosal extension (141); using these radiographic findings, the authors achieved an overall accuracy of 85% for identifying depth of invasion.

Similar to a barium enema, small flat colon adenomas and adenocarcinomas may not be detected with conventional endoscopy because of their similar translucency to surrounding mucosa. They are identified as a slight mucosal deformity, a slightly more reddish color than surrounding mucosa, and by loss of the vascular network pattern. Undoubtedly these small lesions were often previously overlooked.

Potentially, CT colonography with IV contrast will detect flat neoplasms. These tend to be slightly hypervascular compared to normal colonic mucosa and, especially with the higher

resolution available with multidetector CT, should be detectable with a high-quality study.

Linitis Plastica: The rare primary colorectal linitis plastica, or scirrhous carcinoma, usually develops in a setting of inflammatory bowel disease and in younger patients than more typical colon cancers. Metastases are not uncommon when such a primary tumor is first identified.

Both colonoscopic and barium enema findings can be subtle, with a typical appearance resembling a benign stricture (Fig. 5.26). Computed tomography reveals these scirrhous carcinomas as circumferential, homogeneously enhancing lesions. The involved colon wall is thickened considerably. The sensitivity in detecting these lesions depends on tumor size and quality of CT study. Endoscopic US of rectal linitis plastica shows a circumferential thickening of the rectal wall, with thickening involving mostly the submucosa and muscularis propria; endoscopic US also detects perirectal fat infiltration.

In general, breast and stomach carcinomas metastatic to the colon have a linitis plastica appearance more often than a primary colon scirrhous carcinoma.

Figure 5.26. Colon linitis plastica. (Courtesy of Arunas Gasparaitis, M.D., University of Chicago.)

Perforation: Previously performed watersoluble contrast enema has been replaced by CT. Computed tomography readily identifies these mostly advanced tumors, with a majority being associated with an abscess near the tumor.

An occasional carcinoma perforates into adjacent soft tissues and leads to extraintestinal gas or subcutaneous emphysema (Fig. 5.27). At times unusual fistulas form. Technetium-99m- DTPA renography in a patient with hematuria revealed sigmoid colon radioactivity extending to the transverse colon (142); a sigmoid adenocarcinoma had invaded the bladder and formed a colovesical fistula.

Do patients with a perforating colon carcinoma have a worse prognosis than those without a perforation? Comparing perforating cancers and obstructing cancers undergoing emergency surgery, no significant difference in survival or disease progression was evident between these two groups (143).

Obstruction: The size of colon cancers when first detected have decreased during the last several decades, yet it is still common in most practices to see a patient first present with

Figure 5.27. Perforated right colon carcinoma (arrows) in a patient suspected to have acute appendicitis. Soft tissue gas in the necrotic tumor mimics an appendiceal abscess.

ADVANCED IMAGING OF THE ABDOMEN

Figure 5.28. Obstructing carcinoma (arrow). CT colonography can also study the proximal colon. Sagittal images are helpful in surgical planning. (Courtesy of W. Luboldt, M.D., Johann Wolfgang Goethe University, Frankfurt-am-Main.)

colonic obstruction due to a large, bulky tumor. These patients first undergo proximal colon decompression and only later have definitive cancer resection.A sufficiently tight obstruction obviates both a complete barium enema and colonoscopy, studies not only defining an obstructing tumor but also detecting any synchronous neoplasm. In such a setting, preoperative CT colonography is very useful to evaluate the proximal colon. In 19 patients with distal occlusive colorectal carcinomas, preoperative CT colonography identified all occlusive cancers and also detected synchronous lesions—two cancers and 20 other polyps (144), findings confirmed by other studies (145) (Fig. 5.28).

Expandable intraluminal stents are useful in malignant colonic obstructions. A pretherapy stent placed through an obstruction provides decompression, allows a bowel-cleansing regimen to be employed, and thus obviates a preliminary colostomy (146). After decompression, these patients undergo tumor staging, and a decision is made whether to proceed to cancer resection or whether successful stenting is to be the primary palliative therapy. The success rate in stent placement varies but typically is about 90%; thus stent placement was successful in 88% of 80 patients and bowel obstruction

COLON AND RECTUM

resolved in 67% (146). A multicenter study of 71 patients with acute malignant obstruction found self-expandable metallic stent placement to be technically successful in 90%, but it was not possible to advance across the obstruction in 3% and the prostheses was poorly positioned in 7% (147).

Stent complications include perforation and stent dislocation. Completely covered stents tended to migrate more than uncovered stents.

Metastasis as the Initial Presentation: Only an occasional colorectal cancer presents first as a metastasis, generally in the liver. A rare rectal cancer spreads via systemic veins, but pulmonary metastases also occur occasionally from a nonrectal site.

Bone and cerebral metastases as an initial presentation are rare. A cecal carcinoma in a cirrhotic patient first presented with umbilical metastasis (Sister Mary Joseph node) (148). A curiosity is a single microscopic metastatic focus in a resected thyroid colloid nodule in a patient with unsuspected sigmoid colon carcinoma and multiple liver metastases (149).

Staging

General: Several staging systems are in use, including the tumor, node, metastasis (TNM) system (Table 5.4). The Dukes staging system was originally designed for rectal carcinomas, but over the years it has been expanded to include colon cancers; a number of modifications and subdivisions have evolved, and if the Dukes system is used, the specific modification employed should be identified.

A small colorectal cancer initially tends to grow more circumferentially rather than longitudinally along the colon wall. Spread occurs via both the lymphatics and hematogenously. A cancer in the intraperitoneal colonic segments is prone to form peritoneal carcinomatosis once the serosal barrier is breached.

Some carcinomas, especially welldifferentiated ones, invade extensively into surrounding organs without evident metastasis to lymph nodes or more distant structures. For example, a large transverse colon carcinoma in a 60-year-old woman had invaded the adjacent duodenum and pancreas and was in close contact to the superior mesenteric vein (150); no metastases were evident and en bloc resection revealed no lymph node spread.

Table 5.4. Tumor, node, metastasis (TNM) staging of colorectal tumors

AJCC, American Joint Committee on Cancer; UICC, Union Internationale Centre le Cancer.

* Dukes B is a composite of better (T3, N0, M0) and worse (T4, N0, M0) prognostic groups, as is Dukes C (any T, Nl, M0 and any T, N2, M0).

Source: From the AJCC Cancer Staging Manual, 6th edition (2002), published by Springer-Verlag, New York, NY, used with permission of the American Joint Committee on Cancer (AJCC), Chicago, IL.

Ability to detect perirectal node involvement varies with node size. In general, malignant nodes are larger than nonmalignant ones, although some normal sized nodes are invaded and some enlarged ones are not. Enlarged lymph nodes can be due to reactive inflammation. In addition, metastatic nodes range from being partially to totally invaded. Similar to metastatic nodes at other body sites, aside from

node size, node metastases do not correlate with a specific imaging appearance. Also, some colorectal cancers metastasize to more distant lymph nodes and bypass closer nodes. Such skipping nodal metastases are detected in about 10% of patients; patients with skipping nodal metastases have a significantly better prognosis than those without bypassed metastases.

Routine preoperative CT and MR colon cancer staging is of limited use because of low accuracy in assessing the depth of tumor invasion and detecting early lymph node invasion, but these examinations are very useful for detecting invasion of adjacent structures and metastasis to distant sites. Almost all colon cancers are resected. Patients with rectal cancer, on the other hand, have additional therapeutic options, and initial staging often determines the type of therapy employed.

The Radiology Diagnostic Oncology Group concluded in 1996 that CT and MRI accuracies were equivalent in depicting transmural tumor spread, assessing lymph node involvement, and detecting liver metastases (151). Due to advances in CT and MR equipment and software design since then, however, these results should be viewed as obsolete.

Staging is best approached by treating rectal and nonrectal cancers separately.

Rectal Carcinoma: Rectal wall penetration and pelvic lymph node involvement are the major prognostic factors in predicting recurrence. Some lymph nodes <5mm in diameter already contain metastases, a limitation in the imaging prediction of tumor spread. Nevertheless, the sensitivity for detecting positive lymph nodes is greater for rectal tumors than for more proximal colonic tumors because benign perirectal adenopathy is uncommon.

The prevalence of lymph node involvement with rectal cancers is related to tumor depth. Among rectal cancers, lymph node involvement was as follows: T1, 6%; T2, 20%; T3, 66%; and T4, 79% (152). A biopsy finding of lymphatic vessel invasion was highly indicative of lymph node metastasis.

One pathway for the spread of sigmoid and high rectal cancers is via the inferior mesenteric lymph chain, but specific spread is unpredictable and can include the inferior mesenteric lymph nodes, nodes adjacent to rectum, and nodes at the root of the inferior mesenteric artery.

ADVANCED IMAGING OF THE ABDOMEN

A meta-analysis of articles published up to 2002 found that for muscularis propria invasion by a rectal cancer US and MR had similar sensitivities but US specificity was 86% and MR 69% (153); sensitivity for perirectal tissue invasion was: CT 79%, US 90% and MR 82%, with similar specificities. All three modalities were comparable for detecting lymph node involvement.

Multidetector CT is more accurate in staging more advanced rectal cancers than more superficial ones; CT does not provide rectal wall details. Adding multiplanar reconstruction improves local staging of these cancers.

In 53 consecutive patients with distal rectal carcinoma, CT sensitivity for detecting perirectal and inferior mesenteric lymph node metastases was 53% and specificity 85% (154).

Computed tomography using a water enema (hydro-CT) appears useful in staging rectal cancers. Hydro-CT studies tend to be more accurate than no enema studies; increased accuracy is mostly in detecting invasion within or beyond the muscular layers. A CT study of patients with rectal cancer using a tap water enema, IV contrast, and pharmacologic bowel hypotonia reached a sensitivity of 90% and specificity of 70% in differentiating tumors limited to bowel wall from those invading extrinsically (155).

Conventional US has been largely supplanted by endorectal US in staging rectal carcinomas. Endorectal US is very accurate in T-staging superficial cancers but not more advanced cancers because of limited acoustic range. Some authors express endorectal US staging using TNM nomenclature—called the uTNM classification. Although some results are promising, the overall conclusions are rather pessimistic, especially for detecting lymph node metastasis.

Endoscopic resection should be possible if imaging could differentiate between mucosal and submucosal invasion. A number of stuties of early rectal cancer concluded that endoscopic US is not accurate enough to determine appropriate therapy for these tumors. On the other hand, a more recent study found that endorectal US achieved a sensitivity and specificity of 93% and 71%, respectively, and MRI 100% and 60% for detecting rectal wall penetration (156). Endorectal US does not reliably detect muscularis propria invasion (T2 tumors). Endorectal