Материал: Advanced Imaging of the Abdomen - Jovitas Skucas

COLON AND RECTUM

US does, detect perirectal fat invasion (T3 tumors). Tumor spread to more distal structures, such as bladder, is more problematic.

Ultrasonography of the distal rectum is more difficult and the tissue planes are less well defined than more proximally, and endoscopic US staging accuracy in the distal rectum is lower than in the middle or proximal segments. On the other hand, endorectal US is accurate in evaluating anal canal infiltration by low rectal cancers. The overall tendency in staging rectal villous tumors with endoscopic US is to overstage rather than understage. Endoscopic US in patients with a rectal adenocarcinoma performed within 2 weeks prior to surgery and radiation therapy overstaged 21% and understaged 9% of patients (157). Overstaging of the depth of invasion is due, in part, to tumors located close to an uninvolved layer, not uncommon adjacent inflammation and hypervascularity, which tend toward a more anechoic appearance than tumor infiltration. Microscopic invasion accounts for some understaging.

After a rectal polypectomy with an adenocarcinoma discovered in the specimen, endorectal US detected residual tumor with a sensitivity of 100% but a specificity of only 44% (158).

How accurate is endorectal US in detecting local lymph node invasion? Considerably more lymph nodes are involved at histologic examination than are detected by US, and a number of studies have concluded that endoscopic US is too unreliable to be used in preoperative patient selection. Some enlarged lymph nodes are due to reactive inflammation and not tumor infiltration. The ability to detect perirectal nodes varies with node size. In general, malignant nodes are larger than nonmalignant ones, although overlap exists. 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. Several studies concluded that endoluminal US does not reliably identify the extent of lymph node involvement. The role of 3D endorectal US in staging rectal carcinoma remains to be established.

Whether patients are treated by radiotherapy or chemotherapy prior to surgery also appears to influence rectal US accuracy. Diagnostic US accuracy for wall infiltration and lymph node

detection decreases after preoperative radiotherapy or chemoradiotherapy.

In patients with known rectal tumors, endorectal US and MRI achieve comparable staging results. The advantages of endorectal US are its small-diameter instruments, its ready availability, it is technically less demanding, and it costs less; MRI, on the other hand, is operator independent and also evaluates other sites.

Magnetic resonance imaging using an endorectal coil provides more detail than similar MRI using a surface coil. On the other hand, use of an endorectal coil combined with imaging using external coils appears advantageous in assessing both intramural tumor infiltration and more distal pelvic spread. Little data exists on placing such combined imaging in a proper perspective.

High-spatial resolution MRI using T2weighted FSE images identifies mesorectal fascia, peritoneal reflection, Denonvilliers fascia and adjacent structures. Magnetic resonance imaging has a tendency to understage rectal carcinomas (Fig. 5.29). Overstaging in some patients is due primarily to the presence of adjacent perirectal inflammation.

Rectal distension by a water enema improves MR detection of rectal wall penetration. Using a rectal ferric ammonium citrate enema, spin echo MRI identifies most rectosigmoid cancers. Preoperative MR staging of rectal cancer achieved a 100% sensitivity and 70% specificity in distinguishing tumor stages worse than Dukes’ stage A (159); on T1-weighted images a rectal superparamagnetic contrast enema creates a signal void in a distended lumen while wall contrast enhancement by IV gadolinium differentiates mucosa, muscle layers, and perirectal space, details not obtainable with nonenhanced images.

In a randomized phase II trial, preoperative MR using the superparamagnetic iron oxide rectal contrast agent ferristene and IV contrast achieved a sensitivity of 97% and specificity of 50% in staging carcinomas higher than T2 stage (160); using receiver operator characteristic (ROC) analysis, MR differentiated between T1/T2 and T3/T4 tumor stages with a ROC index of 0.85. Higher viscosity rectal contrast agent formulations were superior to lowviscosity formulations, but no significant differences were found between high and low iron content agents.

Similar to other imaging modalities, small lymph node metastases are not identified, although iv SPIO agents show promise. Nonmalignant nodes are hypointense or have a hypointense center, while eccentric and hyperintense nodes tend to contain metastases larger than 1 mm in diameter, although overlap does exist (161).

Can MRI predict sphincter salvage in these patients? A prospective MRI study of patients with a low or middle third rectal adenocarcinoma (defined as <12cm from pectinate line) using rectal and IV contrast and a flexible surface coil achieved 100% sensitivity and 98% specificity in assessing anal sphincter infiltration and 90% sensitivity and 100% specificity in detecting adjacent organ infiltration (i.e., T4 stage) (162); nodal staging, however, was sub- optimal—MRI reached a 68% sensitivity and 24% specificity.

Magnetic resonance imaging potentially differentiates between mucinous and nonmucinous rectal tumors. On T2-weighted fast SE images mucinous tumors had significantly higher tumor-to-muscle, tumor-to- fat, and tumor-to-urine signal intensity ratios compared with nonmucinous tumors (163); most mucinous tumors also revealed peripheral post–gadolinium contrast enhancement.

Nonrectal Carcinoma

Although some earlier studies claimed high CT sensitivity and specificity in detecting local tumor extension (T stage), more recent studies are more pessimistic. Local CT staging accuracy increases at higher disease stages. In general, the presence of obvious pericolic spread and nodal involvement is more reliable than negative findings. Nodal involvement is assessed by simply detecting enlarged nodes. Disappointingly, an accuracy of only about 50% has been achieved by a number of studies for nodal metastases.

Most current CT colonography research involves tumor detection, although some evidence suggests that it also should have a role in staging. Thus one study achieved an overall accuracy of 83% in T staging and 80% in N staging when employing contrast enhanced transverse and multiplanar reformated CT colonography, with images obtained in the arterial phase focusing on the suspected neoplasm and portal venous phase images on the entire abdomen and pelvis (164).

The results of hydrocolonic US to detect and stage colorectal carcinomas have been disappointing. Assessment of lymph node involvement is poor.

COLON AND RECTUM

Comparing different MR studies is difficult because results vary considerably depending on sequences employed. A number of studies have concluded that MR staging accuracy, including nodal involvement, is similar to that obtained with CT. Nevertheless, MRI staging is still in its infancy and future improvements hold promise.

18F-fluoro-deoxy-D-glucose PET appears useful in staging colon cancer. It identifies primary carcinomas (including in-situ carcinomas) but is insensitive in detecting lymph node metastases (165); similar results were obtained with CT. On the other hand, 18F-FDG-PET detects extrahepatic colon metastases missed by other imaging modalities (including CT and MRI); PET detects extraperitoneal nodal metastases, pulmonary metastases, and regional lymph node involvement. Thus PET influences patient selection for hepatic resection.

But 18F-FDG-PET is false-positive in patients with inflammatory bowel or other sites of inflammation.

Radioimmunoscintigraphy holds promise in detecting metastases, but sufficient clinical data are not available on its use in initial staging. Operative gamma probe immunoscintigraphy achieves high sensitivity in detecting liver and extrahepatic abdominal tumor sites but is little practiced.

Therapy

Rather active current research interests involve immunotherapy and genetic therapy. A current trend is toward more aggressive therapy of metastases with such modalities as hyperthermia, cryoablation, and various combination of systemic therapy and surgery. Currently a majority of patients undergo colon cancer resection even in the face of metastases.

For Cure

Endoscopic: A cancer that has not penetrated through the muscularis mucosa is considered a carcinoma-in-situ, and generally endoscopic polypectomy suffices. Terms synonymous with carcinoma-in-situ include intramucosal carcinoma, carcinoma limited to mucosa or lamina propria, and superficial carcinoma. The latter term, however, is sometimes applied to cancers

that also involve the submucosa and thus has an imprecise meaning.

Spread beyond muscularis mucosa signifies an invasive cancer and wider excision is indicated, although some pedunculated polyps are resected endoscopically if no vascular or stalk invasion is evident (Fig. 5.30).

One interesting fusion of procedures consists of laparoscopically assisted endoscopic polypectomy for broad-base polyps. Under general anesthesia, endoscopic polypectomy is assisted by laparoscopy and the polyp site is then sutured using a laparoscopic approach.

Surgical: Colorectal cancer surgery is performed in the very old, with cancer-free survival in these elderly patients being quite good. Indications for surgery should be rather liberal; this is in contradistinction to gastric cancer surgery where survival is much more limited.

In addition to preoperative imaging, liver palpation at laparotomy aids in detecting liver metastases at the time of surgery. Intraoperative liver US is also very useful in further staging and patient selection for additional therapy but currently is not widely practiced. The quality of operative US influences considerably the results obtained.

Preliminary results of laparoscopic colon cancer resection are encouraging, with patient survival, tumor recurrence, and mortality being similar to those of open resection. The liver cannot be palpated during laparoscopic colo-

Figure 5.30. Pedunculated sigmoid adenocarcinoma (arrow). The stalk is seen as a circle within this 1-cm round, smooth tumor.

rectal cancer resection, and this useful diagnostic procedure, performed during an open resection, is thus lost. One proposal is that intraoperative laparoscopic US be included as part of laparoscopic colorectal cancer resection; it is possible to scan all liver segments through a single port site for possible metastases. Nevertheless, this is a complex procedure requiring that a radiologist be present in the surgical suite; all parenchymal segments and major intrahepatic vascular and biliary structures need to be identified during the scan. This takes considerable time and effort.

European clinical trials of rectal cancer patients suggest that best results are achieved with radiation therapy followed by surgery. In the United States postoperative radioand chemotherapy are used for T3 and N1 cancers. One complication after radical surgery for rectal carcinoma is vesicourethral dysfunction.

Radiochemotherapy: Local excision or endorectal radiotherapy are alternate therapies for local control of some select early rectal cancers. A number of studies have established that radiation therapy preand postsurgery improves survival. Preoperative radiotherapy appears superior for local tumor control. Radiochemotherapy alone is appropriate in a setting of an unresectable tumor.

Antitumor activity of some chemotherapeutic agents consists of thymidylate synthetase inhibition, an enzyme necessary in DNA synthesis (166); dose-limiting toxicities prevent their more widespread use. Nevertheless, adjuvant therapy with 5-fluorouracil and levamisole does increase the cure rate of stage III (Dukes’ C) colon cancer patients. Likewise, radiation therapy combined with chemotherapy appears to have a role in patients with stages II (Dukes’ B2) and III rectal cancer.

A combination of adjuvant radiochemotherapy and radical surgery in patients with rectal carcinoma achieves mixed results. Complete response to preoperative chemotherapy and radiation therapy for locally advanced rectal cancer is achieved in a minority. In some patients such therapy also decreases cancer stage, thus permitting a sphincter-saving procedure.

Computed tomography has a major role in planning preoperative radiotherapy.

Serial postcontrast infusion MRI in patients with T3 rectal carcinoma during preoperative

ADVANCED IMAGING OF THE ABDOMEN

radiotherapy provides tumor perfusion data from which a tumor perfusion index (PI) was established (167); the PI increased significantly during the first 2 weeks of therapy, and then decreased. A high initial PI value correlated with subsequent greater lymph node downstaging and thus is potentially of prognostic significance.

Radioimmunotherapy: The use of monoclonal antibody radioimmunotherapy for primary colorectal cancer is still in investigational status. Simultaneous injection of copper 67 and sodium iodide 125–anti-CEA monoclonal antibody in six patients yielded an average Cu-67/I-125 ratio of 1.9 for tumor uptake, 0.7 for blood, and 2.6 for tumor to blood (168); one problem identified by the study was that Cu 67–monoclonal antibody tumor uptake was too low while liver and bowel uptake was considerable.

Other: Ten patients with advanced pelvic cancer (recurrent rectal and ovarian cancer) underwent sequential arterial cisplatinum and mitomycin infusion via an extracorporeal circuit established by isolating pelvic vessels with balloon catheters placed above the aortic and caval bifurcations and pneumatic cuffs at the thighs (169); although the authors established the feasibility for such extracorporeal perfusion, 2-year patient follow-up failed to show a positive response.

Palliation

In the absence of metastases, a preoperative definition of an unresectable cancer is often imprecise. Especially for rectal cancers, unresectability is often established only at surgery.

Patients with unresectable rectal cancer undergo palliation therapy. A combination of local excision, radiation, and chemotherapy is balanced depending on the size of the local tumor spread, the size of the metastases, and life expectancy. Repeat local excision is appropriate for some. Even pelvic reradiation appears to have a role in select patients. Published reports of brachytherapy have been disappointing.

Endoscopic laser therapy using an Nd:YAG noncontact laser performed in patients with obstructing or bleeding inoperable colorectal cancers initially control symptoms in most patients, but symptoms can be expected to recur.

COLON AND RECTUM

For Obstruction: Most stents are inserted retrograde for palliation of rectosigmoid tumors. An occasional one is inserted antegrade via a percutaneous cecostomy. In addition to relief of obstruction prior to surgery, expandable intraluminal stents decompress the bowel for palliation of a nonresectable tumor, obviating a permanent colostomy (146). A multicenter study of successful palliative stent placement found bowel obstruction resolving within 24 hours of stenting in 96% of patients, with none requiring a colostomy for decompression (170).

Complications, consisting of mild rectal bleeding, abdominal pain, stent malpositioning, obstruction due to fecal impaction, and eventual tumor ingrowth into the stent lumen, are not uncommon. Stents have perforated and migrated.

Among survivors, estimated primary stent patency rate was 91% at 6 months (146).

For Bleeding: An occasional patient with colorectal cancer presents with massive and potentially life-threatening bleeding. Even in a setting of an unresectable tumor, transcatheter embolization should be encouraged. Reembolization is performed if bleeding recurs. Most of these patients eventually die from tumor cachexia rather than exsanguinate.

Recurrence and Follow-Up

General: Although screening for recurrence is commonly practiced, guidelines on specific follow-up have not been established. Screening practices for hepatic and pulmonary metastases vary considerably between hospitals.

Currently a CEA determination is the most common test used to detect cancer recurrence. Carcinoembryonic antigen has a high specificity for tumor recurrence and not uncommonly is positive before imaging. On the other hand, whether frequent CEA determinations prolong survival is not clear.

The role of colonoscopy in screening for local recurrence is rather limited because only about 10% of recurrences are intraluminal. Colonoscopy does have a long-term role, however, in detecting new adenomas and metachronous cancers. Although colonoscopy does have a long-term role in detecting new adenomas and metachronous cancers, preliminary studies suggest, however, that CT colonography

is a feasible alternative to both conventional colonoscopy and liver US in following these patients (171).

Both CT and MRI have a major role in detecting recurrence of colorectal cancers, with some studies claiming an accuracy over 90%. A basic question, however, is whether such early recurrence detection influences survival. Current data suggest a rather pessimistic answer. A prospectively, randomized study with patients undergoing either intensive (yearly colonoscopy, liver CT, chest radiography, clinical review, and simple screening) or standard follow-up (structured clinical review and simple screening tests only) found no significant difference in survival between the two groups after a 5-year follow-up (172); yearly colonoscopy failed to detect any asymptomatic local recurrence, and only one asymptomatic curable metachronous colon tumor was found. Liver CT resulted in earlier detection of hepatic metastases but did not increase the number of curative hepatectomies.

Local peritoneal involvement appears to supersede other parameters in estimating patient prognosis. Nevertheless, the clinical significance of malignant cells in the peritoneal cavity is not clear. For instance, in patients with no evidence of peritoneal metastases, peritoneal washing before elective colon resection for adenocarcinoma detected malignant cells in 32% of patients with tumor extending to the serosa (173); the 5-year survival rates were 48% for those with positive washing and 68% for those with negative washing, although multivariate analysis revealed no significant association between positive washing and survival.

Rectal Carcinoma: Recurrence of rectal carcinomas is discussed separately because its recurrence patterns and imaging approach are distinct from those in more proximal colon. Tumor recurrence must be distinguished from postresection fibrosis and surgical deformity (Fig. 5.31). Staging of patients with a proven rectal cancer after radiation and chemotherapy is especially difficult. Both CT and endorectal US overstage these cancers even in patients subsequently found to have no residual cancer. No current imaging modality can reliably distinguish between radiation fibrosis and residual cancer, although MR and scintigraphy show promise (see below). With most imaging a baseline study 2 to 4 months after surgery is

Figure 5.31. A double-contrast barium enema reveals a benign anastomotic stricture (arrow) after sigmoid resection for cancer.

very useful. Postoperative fibrosis retracts over time and gradually develops well-defined margins.

Recurrence after an abdominoperineal resection is most often local. It tends to infiltrate diffusely and is asymmetrical in appearance (Fig. 5.32). No consensus exists on the best imaging

ADVANCED IMAGING OF THE ABDOMEN

modality to evaluate for recurrence. Recurrence detection rates of >90% by endorectal US are being achieved, and this study appears especially useful in this setting.

CT achieves about an 80% sensitivity and specificity in detecting recurrence, while MRI sensitivity is similar but specificity is greater. A typical protocol for these patients consists of CT performed within 2 to 4 months after resection and repeated every 6 to 8 months during the first 2 years (with CEA testing); MRI is reserved for those with a positive or questionable CT finding, those with different clinical symptoms, and those with an increasing CEA. A biopsy is indicated if MRI does not resolve the issue.

Postcontrast CT of active granulation tissue reveals considerable enhancement. Compounding the issue is that local recurrence of rectal cancer results in transient early enhancement at the anastomotic site; whether recurrence can be confidently differentiated from postoperative granulation tissue is not clear.

After rectal cancer resection, the most common CT finding of recurrence is a round or nodular enhancing tumor. A retrospective study of postsurgery or radiotherapy rectal cancer patients who had at least three CT examinations concluded that relapse should be considered if a presacral mass enlarges, appears inhomoge-

COLON AND RECTUM

neous, and is asymmetric in outline, or if enlarged lymph nodes develop and infiltrate surrounding structures (174); an unchanged appearance in several follow-up CT examinations is evidence for lack of recurrence. A necrotic recurrence is often difficult to distinguish from an inflammatory mass or abscess.

Some evidence suggests that radiation fibrosis has a somewhat different MR enhancement pattern than postoperative scar tissue. Granulation tissue developing shortly after rectal cancer resection shows marked postcontrast CT enhancement and is hyperintense on T2-weighted MR images. Eventual fibrosis, especially evident after radiation therapy, is hypointense on both T1and T2-weighted images. Fibrosis shows poor but variable enhancement postcontrast. Retraction of surrounding tissues is common. Fibrosis exhibits an irregular enhancement even years after radiation therapy.

Technetium-99m–labeled anti-CEA antigenantibody scintigraphy appears to have a role in detecting pelvic recurrence, with sensitivity being similar to that of CT. Sensitivity is greater with an increase in size of a recurrence. In select patients antibody scanning aids in differentiating recurrent tumor from fibrosis.

18F-fluoro-deoxy-D-glucose PET shows increased tumor uptake at recurrence sites; scar tissue has low FDG accumulation. FDG-PET in one study achieved a sensitivity of 82% and a specificity of 65%, but combined PET/CT sensitivity increased to 98% and specificity 96% in differentiating malignant from benign disease (175).

With suspected local recurrence, either transrectal-guided fine-needle aspiration cytology using a 21-gauge needle or a core biopsy with an 18-gauge needle appears appropriate. If a tumor is palpable, digitally guided puncture is feasible; for others, either CT or US guidance is used. Whether 3D endorectal US guidance aids the biopsy of suspected perirectal recurrence remains to be established.

Nonrectal Carcinoma: As a rough estimate, approximately 50% of patients undergoing curative resection for Dukes C colorectal cancer develop a recurrence within 3 to 5 years, and over 90% of them die from their cancer. In those developing a recurrence, almost all colorectal cancers recur within 3 years. Because many recurrent carcinomas consist primarily of an

extrinsic component, CT and MRI are well suited for postresection follow-up. A baseline study aids the future differentiation of fibrosis from recurrent tumor.

Superficially, a suture granuloma developing after bowel resection and reanastomosis mimics local recurrence with both colonoscopy and a barium enema.

Considering enhancement within 90 seconds of an abnormal structure on dynamic contrastenhanced subtraction MRI to signify a malignancy, MR achieved a 97% sensitivity and 81% specificity in differentiating fibrosis from recurrence during follow-up (176); using a finding of high signal intensity on T2-weighted SE images as a criterion for malignancy, sensitivity and specificity were only 77% and 56%, respectively.

Distal Recurrence: For unknown reasons, colorectal cancer metastasis to a fatty liver is uncommon.

Colorectal cancer metastasis to bone is not uncommon. At times bone biopsy is required both for diagnosis and to exclude osteomyelitis. Bone metastasis is more common with rectal and cecal cancers than with other colon cancers. What is uncommon is to find a solitary bone metastasis or a metastasis years later. Signet ring cell cancers appear to have a higher propensity for bone metastasis. Brain metastasis is uncommon but does occur. A rare metastasis occurs within a laparotomy scar.

In women with a prior colorectal adenocarcinoma who then developed a new pelvic tumor, ovarian metastasis was found in 57%, a benign ovarian neoplasm in 26%, and a primary ovarian cancer in 17% (177); among women with a past colorectal cancer, a newly diagnosed uterine cancer was a primary endometrial adenocarcinoma in 73% and metastatic colon cancer in 20%. The rare colorectal carcinoma in adolescent girls frequently metastasizes to the ovaries; these ovarian metastases range from solid, combined solid and cystic, to multilocular cysts, although they tend to be cystic less often than in older patients.

Calcifications in metastatic colon carcinoma are not uncommon; such calcifications are readily identified with CT. Ossification in a metastatic colon carcinoma is rare.

Imaging studies useful in evaluating liver and extrahepatic metastases include conventional chest radiographs, CT, MRI, and scintigraphy.