The Large Bowel

Anatomy

Detailed anatomical knowledge of the large intestine is fundamental to accurate image interpretation. The large bowel comprises the colon, vermiform appendix, rectum and anus. The caecum, ascending and descending colon are covered anteriorly by visceral peritoneum, whereas approximately 50% of the posterior aspect of these segments is retroperitoneal. The retroperitoneal colon has an adventitial layer, separating muscle from peritoneal fat. The anterior peritoneum runs medially onto the rudimentary mesocolon and laterally onto the abdominal wall as parietal peritoneum. The transverse and sigmoid colon have a mesentery formed from a double layer of visceral peritoneum sandwiching connective and adipose tissue with vessels, nerves and lymphatics. Other colonic segments may occasionally also have a mesentery as a normal variant. The outer colonic muscularis propria has two layers, an inner circular and an outer longitudinal, with the myenteric (Auerbach) nerve plexus between. The outer layer is thin, except where it is condensed into three narrow bands called the taeniae coli that contain more collagen and elastic tissue than muscle. Three rows of haustral sacculations arise between the taeniae, with haustral folds between the sacculations. Distal colonic haustra form only when the taeniae contract. The intraperitoneal colon is covered by mesenteric serosa. Subserosal caecal and sigmoid fat accumulates in small peritoneal pouches to form the epiploic appendages that may encase diverticula in the sigmoid. The superior mesenteric artery supplies the colon proximal to the splenic flexure via the ileocolic, right and mid-colic branches. The colon distally is supplied by the left colic, sigmoid and superior rectal branches of the inferior mesenteric artery. The marginal artery is a vascular arcade running along the mesocolonic border formed from the terminal branches of the superior and inferior mesenteric arteries. It gives off short branches, the vasa recta, which penetrate the muscle layer close to the taenia mesocolica, and long branches entering between the taenia omentalis and libra. The veins and lymphatics follow the course of the arteries, draining into the portal vein and coeliac nodes, respectively. The mid-rectal veins drain via the internal iliac veins into the systemic circulation through the inferior vena cava. From the rectum, lymph drains superiorly via superior rectal artery nodes to the inferior mesenteric chain, posteriorly by nodes around the median sacral artery, and laterally around the middle rectal artery to the internal iliac chain.

The rectum is defined as commencing at the third sacral level, although the sacral promontory is often taken surgically as the reference point. Others define it as the distal 15 cm of large bowel proximal to the anus. Anteriorly the rectum is covered by peritoneum to the level of the junction of the upper two-thirds and lower one-third. The lateral and posterior aspects of the upper rectum and all the lower one-third are surrounded by the mesorectum, which is composed of loose adipose connective tissue containing the small perirectal lymph nodes and the superior rectal vessels. The mesorectum itself is enclosed by the mesorectal fascia ( Fig. 22.1 ). Posteriorly the mesorectal fascia is separated from the presacral fascia by the thin retrorectal space; anteriorly it blends with the urogenital septum (Denonvillier fascia/rectovaginal septum); superiorly it is contiguous with the sigmoid mesentery; and inferiorly it terminates close to the anus in the parietal fascia covering the levator ani. There is no rectal haustration, but the rectal valves (of Houston) create mural folds. The anus has a complex sphincter arrangement with an internal sphincter comprised of smooth muscle (a continuation of the circular muscle coat of the distal rectum) and an external sphincter of striated muscle. Between these, the longitudinal layer, consisting of striated and smooth muscle with extensive fibroelastic tissue, anchors the anus in position. The vascular subepithelial tissues help seal the canal to maintain continence.

The pelvic floor is an anatomical and functional unit that consists of muscles and connective tissue in three contiguous supporting layers. From cranial to caudal, these include the fascial layer or endopelvic fascia, the intermediate pelvic diaphragm and the urogenital diaphragm. The muscular pelvic diaphragm consists mainly of the levator ani complex.

The colonic mucosa is columnar in type with goblet and some enterochromaffin cells arranged in crypts (of Lieberkühn). The surface pattern consists of fine parallel grooves running transversely with short intercommunicating branches called the innominate groove pattern. The lamina propria contains lymphoid follicles, the submucosa, adipose tissue with neural elements (Meissner plexus), blood vessels and lymphatics.

Radiological Investigation

The mainstays of colonic radiological investigation are intraluminal contrast examinations and cross-sectional techniques. Although cross-sectional imaging is largely replacing contrast studies, the latter still maintain a role in specific clinical situations.

A water-soluble contrast enema (e.g. Gastrografin diluted 3:1 with water or Urografin 150; Schering AG, Berlin, Germany) allows real-time evaluation of colonic anatomy and is most commonly used to test the integrity of surgical anastomoses ( Fig. 22.2 ), define colonic calibre (e.g. in suspected megacolon), delineate colonic fistulae ( Fig. 22.3 ) and exclude mechanical obstruction. These agents are hypertonic and produce diarrhoea, which may be an intended therapeutic effect. Proximal to colonic strictures, fluid shifts will distend the colon and consequent perforation has been reported. Care must be taken not to dilate the bowel proximal to a stricture; hence isotonic contrast agents are increasingly used.

A double-contrast barium enema (DCBE) involves full bowel preparation, an intravenous (IV) smooth muscle relaxant, partial filling of the colorectum with a barium suspension and insufflation with air or carbon dioxide for luminal distension. The objective is to acquire a series of images so that the entire colon is seen in double contrast, with no segment coated poorly or obscured by a barium pool. The test is now obsolete for detection of polyps or cancer, having been replaced by computed tomography colonography (CTC).

CT can be performed either with (CT colonography) or without gaseous colonic insufflation – in general, gaseous distension improves diagnostic accuracy for colonic abnormalities, although it is more invasive and not always appropriate in acute situations. Standard abdominal CT protocols may include colonic opacification using orally-administered contrast (e.g. 2% barium suspension or Gastrografin), typically commencing the day before the examination, together with IV contrast medium administration. Rectal contrast agents may be used, particularly for suspected perforation or anastomotic leak.

The colonic wall should be no more than 3 mm thick when adequately distended. The pericolic fat should be homogeneous with only a few vascular channels. The normal colonic mucosa enhances after IV contrast agent administration, but the pattern of mural enhancement in an abnormal colon may provide clues as to the underlying diagnosis ( Table 22.1 ).

CT colonography (CTC) describes CT of the gas-distended colon. The use of oral contrast agents to ‘tag’ residual colonic contents so that they yield higher attenuation is mandatory and can be combined with full or reduced laxative preparation, or without any additional laxative (‘prep-less CTC’). High osmolar contrast agents, such as Gastrografin, have a laxative effect and can be used as a single colonic preparatory agent. Colonic distension should be performed using carbon dioxide; automated insufflation devices give superior distension to manual techniques. Furthermore, distension is improved by IV spasmolysis using hyoscine butylbromide (Buscopan, Boehringer, Ingelheim, Germany) and dual positioning with supine and prone acquisitions (or decubitus views, if prone positioning is not possible); this is essential to ensure full mucosal visualisation. In some patients, demonstration of the entire endoluminal surface requires supine, prone and decubitus positioning . Interpretation is performed using a combination of two-dimensional (2D) axial and multiplanar projections, together with three-dimensional (3D) endoluminal reconstructions. Cross-correlation between 2D and 3D images is required to differentiate normal colonic structures such as the ileocaecal valve ( Fig. 22.4 ), haustral folds and faecal residue from pathology . Collimation with multidetector CT is typically 1 to 2.5 mm. IV contrast administration is not required for colonic evaluation but is regularly employed to facilitate evaluation of extracolonic disease ( Table 22.1 ).

Magnetic resonance colonography (MRC) follows similar principles to CTC, and is most commonly performed after bowel purgation, although non-laxative approaches are possible. Colonic distension is achieved with 1.5 to 2 L warm water or gas (carbon dioxide or air). Bright lumen MRC uses a gadolinium-spiked water enema and typically a 3D T ₁ weighted spoiled gradient echo (GRE) sequence is used. Dark lumen MRC uses air, carbon dioxide or water and is more widely performed, although it requires the administration of IV gadolinium.

Evacuation proctography (EP) is a study of the dynamics of rectal evacuation. Conventionally the procedure has been performed using X-ray fluoroscopy, but magnetic resonance imaging (MRI) proctography has gained increasing acceptance. The rectum is distended using thick barium paste (fluoroscopy) or air/ultrasound jelly (MRI). During the fluoroscopic procedure, a video recording is made of the voluntary evacuation of the paste. The small bowel should be opacified with a dilute barium suspension to show any enterocele, and some advocate contrast filling of the bladder and vagina. No such extracolonic organ opacification is required for MRI proctography and evacuation is captured using a rapid dynamic sequence such as true fast imaging with steady-state precession (TrueFISP). Proctography may be viewed in three stages: Rest, evacuation and recovery. At rest, the anorectal junction is normally just above the plane of the ischial tuberosities. Evacuation is initiated by descent of the pelvic floor, widening of the anorectal angle, and relaxation of the anal sphincters. The rectum distal to the main fold is squeezed by raised intra-abdominal pressure against the levator ani to form a ‘zone of evacuation’ that typically empties in less than 30s. During MRI proctography, organ prolapse is conventionally measured with respect to the pubococcygeal line ( Fig. 22.5 ) which provides a convenient, reproducible point of reference.

Abdominal ultrasonography requires a graded compression technique for good views of the colon. Intraluminal gas often prevents demonstration of both walls, but the haustral pattern of the anterior wall should still be seen. Usually only the low reflective muscularis propria, reflective submucosa, and hypoechoic mucosal layer are identified ( Fig. 22.6 ). The bowel wall thickness may be measured, Doppler flow assessed and the pericolic tissues interrogated. Endosonography allows higher frequency probes to be used (10 to 20 MHz range) to show the wall layers in detail. The sonographic pattern is created by a mixture of interface reflections between, and reflections from, the thin layers. A four-layer pattern is seen in the anal canal ( Fig. 22.7 ), with a five-layer pattern in the rectum ( Fig. 22.8 ).

Nuclear medicine isotope studies have a relatively limited role in large bowel imaging with the exception of positron emission tomography (PET), usually combined with either CT (PET-CT) or MRI (PET-MRI) for assessment of malignancy. In particular, 18-FDG-PET (radiolabelled 18F-fluorodeoxyglucose) fusion studies are valuable for confident exclusion of extraluminal disease that may remain undetected with routine cross-sectional techniques, particularly where radical surgery may be inappropriate in the context of distant metastases. Colonic cancer and larger adenomatous polyps are often 18-FDG avid and may be found incidentally during CT (PET-CT) or MRI (PET-MRI) performed for non-colonic indications.

Tumours

Polyps

A polyp is an elevated mucosal lesion ( Table 22.2 ). The majority occur sporadically in the general population, although there are many rare polyposis syndromes. The most clinically significant polyps are adenomas. By definition, adenomata contain dysplasia (i.e. intra-epithelial neoplasia). Colorectal cancer (CRC) represents an extension of this neoplasia beyond the muscularis mucosae into the submucosa. Perhaps two-thirds of CRCs originate via an adenomatous precursor (sometimes called the adenoma-carcinoma sequence). Detection of polyps is therefore clinically important, particularly as their removal substantially reduces subsequent carcinoma risk. This is particularly relevant for higher-risk ‘advanced’ adenomas (those with greatest chance of progression to invasive cancer), which are traditionally defined as adenomas that are over 1 cm in diameter, or contain high-grade dysplasia or have greater than 25% villous morphology at histopathology.

TABLE 22.2

Classification of Polyps and Polyposis Syndrome

Histological Type	Single or Few in Number	Polyposis
Epithelial	Adenoma—tubular, villous, tubulovillous	Familial adenomatous polyposis, Turcot syndrome
	Adenocarcinoma	Cowden disease
Hamartomatous	Juvenile	Juvenile polyposis
	Metaplastic	Peutz–Jeghers syndrome metaplastic polyposis
Inflammatory	Post-inflammatory polyp	Post-inflammatory polyposis
Nonepithelial	Lipoma, carcinoid, GIST, benign lymphoid, neurofibroma	Lymphomatous polyposis, metastatic, neurofibromatosis
Miscellaneous	Endometriosis	Cronkhite–Canada syndrome

GIST, Gastrointestinal stromal tumour.

Adenoma prevalence depends largely on age, gender and family history. They are rare under the age of 30 but are seen in around 30% of individuals over 50 years. They are commoner in men than women and a first-degree relative with CRC confers about 50% additional risk. Around 50% occur in the rectosigmoid, 25% in the descending colon, 10% in the transverse and 15% in the caecum and ascending colon with the prevalence of right-sided adenomas increasing in older subjects.

Certain factors that imaging depicts can help estimate the likely clinical significance of a polyp (i.e. the risk of malignancy, both current and future). Size is most important; the risk of invasive cancer in a series of over 11,000 adenomas was negligible for <5 mm polyps, 2% for polyps between 0.6 and 1.5 cm in diameter, 19% for 1.6–2.5 cm polyps, 43% for 2.6–3.5 cm polyps and 76% for polyps of >3.5 cm. Morphology is also important and should be described according to the Paris classification ( Fig. 22.9 ) as this facilitates comparison with the endoscopic literature. Pedunculated polyps (Paris Ip), which have an elongated stalk, are more likely to contain severe (high grade) dysplasia or foci of invasive cancer. On the other hand, because the stalk provides some distance between the epithelium and the bowel wall, they are often considered cured once resected if the neoplasia does not extend beyond the resected polyp's stalk (Haggitt levels 1, 2 or 3). Sessile polyps (Paris Is) have a similar diameter at their base and near the top they are roughly hemispheric. These have an intermediate risk of invasive malignancy for a given size. Subpedunculated polyps (Paris Isp) are intermediate between sessile and pedunculated polyps, both in appearance and risk. ‘Flat polyps’ are not truly polypoid and hence are probably better termed ‘flat lesions’ or ‘non-polypoid colorectal neoplasms’. A flat lesion (Paris 0-II group) is defined as being less than 2.5 mm in height above the colonic mucosal surface at endoscopy, or 3 mm on radiology. Flat lesions may be depressed rather than elevated relative to the colonic mucosa (Paris 0-IIc). These carry a higher risk of invasive cancer for a given size, whereas other subtypes of flat lesions (Paris 0-IIa: slightly elevated; Paris 0-IIb: completely flat) are generally less aggressive. They can be very challenging to detect, both at endoscopy and on imaging ( Fig. 22.10 ).

Non-adenomatous polyps were once felt of lesser clinical importance due to a lower risk of malignancy, although this perspective is evolving. Traditionally, polyps were divided into adenomas, hyperplastic polyps and others (including juvenile, inflammatory, lymphoid and other rare polyps). Some polyps, which were historically classified as hyperplastic, have microscopic architectural distortion and are now termed sessile serrated adenomas. Their genes can show errors in sequences of repeated DNA, termed microsatellite instability (MSI) and/or gene hypermethylation (CpG island methylator phenotype, CIMP). They feed into a second major pathway of colorectal carcinogenesis termed the ‘serrated pathway’, which is hypothesised to account for up to 35% of CRC, particularly in the right colon. This implies polyps other than conventional adenomas may have clinical significance. Furthermore, even true hyperplastic polyps may harbour some genetic alterations also seen in CRC and share many risk factors with adenomas and CRC. They may represent a marker that an individual is at higher risk of subsequent CRC or genuinely represent precursor lesions.

Polyposis Syndromes

Familial adenomatous polyposis.

Familial adenomatous polyposis (FAP) is caused by a mutation of the Adenomatous polyposis coli (APC) tumour suppressor gene on chromosome 5q21, and accounts for about 1% of CRC. Inheritance is autosomal dominant. APC is a critical gene in sporadic carcinogenesis: Individuals with FAP are therefore predisposed towards tumourigenesis. More than 100 adenomas have to be present for the diagnosis; typically, several hundred polyps are present. These may cause rectal bleeding, diarrhoea and mucus discharge. All affected patients eventually develop CRC. Preventative proctocolectomy is therefore recommended.

Extracolonic polyps also occur in this condition: gastric adenomas and hamartomas are present and almost 100% of FAP patients have duodenal adenomas clustered around the ampulla (possibly due to the co-carcinogenic effect of bile). There is a 5% risk of periampullary duodenal carcinoma, a major cause of death in those who have had proctocolectomy. Extraintestinal manifestations include multiple osteomas of the skull and mandible, epidermal cysts (often facial), congenital hypertrophy of retinal pigment epithelium, abnormal dentition and desmoid tumours. Gardner syndrome is a variant of FAP with prominent skeletal and skin manifestations. Desmoid formation is often precipitated by trauma or surgery. These benign fibromatous tumours are locally invasive and involve the abdominal wall, small-bowel mesentery or retroperitoneum. Previously, desmoid disease was a major cause of morbidity and mortality, although this has improved with advances in surgical expertise. In the early stages CT and MRI show ill-defined mesenteric infiltration with small-bowel tethering ( Fig. 22.11 ), before mass development (which may be huge). Imaging often underestimates the extent of disease at surgical resection.

Hereditary non-polyposis colorectal cancer (Lynch syndrome).

Hereditary non-polyposis colorectal cancer (HNPCC) is an autosomal dominant condition caused by faults in DNA mismatch repair (MMR) genes and probably accounts for 5% of all CRC. The lifetime risk of CRC in HNPCC is 70% to 85%. Genetically, these tumours exhibit high levels of MSI. Other tumours are also increased, notably of the endometrium, small bowel and renal pelvis/ureter. Clinical criteria for this condition follow the ‘3-2-1 rule’: (1) three or more relatives with a HNPCC-associated cancer; (2) two or more successive generations affected; (3) one or more tumours diagnosed before the age of 50 years; and (4) one should be a first-degree relative of the other two. Cancers occur at an earlier age in HNPCC (mean 45 years); about 70% are in the proximal colon, and multiple tumours are common. Small polyps at an early age may be present.

Peutz–Jeghers syndrome.

This autosomal dominant condition is characterised by mucocutaneous pigmentation and intestinal hamartomas, mainly in the stomach and small bowel. Large-bowel polyps are fewer, but are larger, often pedunculated, and may bleed. There is an increased risk of gastrointestinal tract cancer, including CRC. Extraintestinal cancers are also increased, particularly of the ovary, cervix, thyroid, testis, pancreas and breast.

Rare polyposes.

There are a number of other rare polyposes. Turcot syndrome is an association between colorectal adenomas/carcinomas and primary brain tumours. It can arise via a HNPCC-type defect in mismatch-repair genes or FAP-type defect in the APC gene. Sebaceous gland tumours occurring with HNPCC-type tumours is called Muir–Torre syndrome. Cowden syndrome is an autosomal dominant trait with hamartomatous intestinal polyposis and lesions of the skin, mucous membranes, breast, thyroid and dysplastic cerebellar gangliocytoma. Juvenile polyposis is also autosomal dominant and presents in infancy with multiple juvenile hamartomatous polyps in the stomach, small bowel and colon. CRC risk is increased Cronkhite–Canada syndrome, which is a diffuse intestinal polyposis associated with alopecia, hyperpigmentation of the skin and nail atrophy secondary to gross malabsorption. Serrated polyposis (previously known as hyperplastic or metaplastic polyposis) is a very rare condition with increased CRC risk and is diagnosed when there are multiple or large, proximally located, serrated colonic polyps.

Radiographic Features of Polyps

Computed tomography colonography.

CTC is the most accurate radiological technique for polyp detection, surpassing DCBE and approaching that of colonoscopy for larger polyps. The entire colonic surface must be evaluated using either a primary 2D approach with 3D problem-solving or a primary 3D endoluminal fly-through with 2D review of potential lesions ( Fig. 22.12 ). The exact analysis method used is less important than the training and experience of the reporting individual. Computer-aided detection (CAD) increases sensitivity with only a small reduction in specificity and should be used as a second reader to maximise its benefit ( Fig. 22.13 ). Once a polyp candidate is detected, it must be interrogated further to confirm its nature. Faecal residue may mimic a polyp; variable attenuation due to internal gas content is a distinguishing feature from the usual homogeneous soft-tissue attenuation of a polyp ( Fig. 22.14 ). Faecal tagging (use of oral contrast agent to label or ‘tag’ residual colonic contents) improves both sensitivity and specificity; lesions which might be obscured by retained liquid residue can be seen within the higher-density fluid ( Fig. 22.15 ) and tagged stool will not be mistaken for a polyp due to its high attenuation. Stool often moves between the prone and supine acquisitions while a fixed mural lesion does not. However, the mesenteric colon may be mobile and hence changes in colonic position between the supine and prone acquisition may suggest ‘movement’ of a genuine polyp; correlation with fixed landmarks such as diverticula or haustral folds may be helpful. Lipomata may be polypoid on 3D but the fat density on 2D is diagnostic ( Fig. 22.16 ). Similarly, while an inverted diverticulum may simulate a polyp on 3D, gas content on 2D review indicates its true nature. Flat lesions can be very hard to detect and may only manifest as subtle wall thickening on 2D images (often best appreciated on an abdominal window setting) or minor protuberance or irregularity on 3D images ( Fig. 22.17 ). Iodinated contrast often adheres to flat lesions which aids conspicuity ( Fig. 22.18 ). CAD can help detect flat lesions and newer CAD algorithms will improve performance further.

Magnetic resonance colonography.

Dark-lumen sequences depict polyps as enhancing protrusions from the normal mucosa ( Fig. 22.19 ). Endoluminal projections are feasible but may appear pixelated due to the current lower spatial resolution of MR.

Double-contrast barium enema.

As the evidence base for CTC evolves, DCBE declines and is becoming increasingly obsolete.

Colorectal Cancer

Primary CRC is the third commonest cancer in the UK with around 42,000 new cases each year (Cancer Research UK, 2016). With 16,000 deaths annually, it is the UK's second commonest cause of cancer mortality. The lifetime risk is 1 in 14 for men and 1 in 19 for women. The risk increases with age, with almost three-quarters of cases seen in people aged 65 or more. Family history is a factor in around 20% of cases of CRC. Other risk factors include obesity, alcohol misuse, cigarette smoking and chronic inflammatory bowel disease (IBD). Conversely, aspirin–in addition to diets low in red meat and high in fibre–is considered to be protective.

It is generally accepted that just over half of all CRCs arise in the rectum or sigmoid, with the rectum alone accounting for one-third of cases. However, the advent of screening, increased polypectomy rates and the aging population is leading to slight changes in this distribution. Overall 5-year survival is about 50%. Major prognostic factors include local tumour stage, vascular or lymphatic invasion, preoperative elevation of carcinoembryonic antigen (CEA) and tumour differentiation (grade).

Colon cancer and rectal cancer are currently treated slightly differently; the mainstay of therapy for both is surgical excision, but in the rectum it is more difficult to achieve adequate clearance margins to prevent local recurrence whilst avoiding significant complications. However, the static nature and pelvic position of the rectum make it amenable to chemoradiotherapy, which has been shown to decrease local recurrence in later-stage disease. Local staging is therefore particularly important in rectal cancer and will likely become increasingly so for colon cancer should current large-scale trials confirm the utility of preoperative chemotherapy for locally advanced disease. Both colon and rectal cancers can be staged according to the Dukes and TNM (tumour, nodes, metastasis) systems ( Table 22.3 ).

TABLE 22.3

Colorectal Cancer Staging (TNM 8th Edition)

UICC/TNM	Tumour Extent	Dukes	5-Year Survival
Stage I	Invasion submucosa T1	A	85%–95%
	Invasion muscularis propria T2
	No nodal involvement, no distant metastasis
Stage II	Invasion outside muscularis propria T3	B	60%–80%
	Invasion visceral peritoneum T4a
	Invasion other organs T4b
	No nodal involvement, no distant metastasis
Stage III	1–3 lymph nodes involved N1	C	30%–60%
	>3 N2
Stage IVa	Distant metastasis in one organ M1a	D	<10%
Stage IVb	Distant metastasis in >1 organ M1b	D	<10%
Stage IVc	Metastasis to the peritoneum with or without distant organ involvement	D	<10%

TNM, Tumour node metastasis; UICC, Union for International Cancer Control (formerly International Union Against Cancer).

Colon Cancer

CTC has equivalent sensitivity to colonoscopy for detecting CRC and recent data suggest the rate of interval cancer after CTC is similar to that of colonoscopy, although one advantage of colonoscopy is the ability to obtain biopsy confirmation of malignancy. CTC readily depicts annular or semi-annular lesions with shouldered ends and luminal narrowing ( Fig. 22.20 ). Similarly, flat or plaque-like cancers show wall thickening and irregularity ( Fig. 22.21 ). At conventional CT, malignancy is usually seen as an area of focal wall thickening (>3 mm), often with extension into the pericolic fat and local nodal enlargement ( Fig. 22.22 ). Tumours are usually homogeneous but may be heterogeneous in the context of large adenocarcinomas or mucinous tumours or when associated with abscess formation. Mucinous tumours, both primary and metastatic, also have a propensity to calcify. Colonic cancer is usually intensively 18-FDG avid on PET-CT ( Fig. 22.23 ), but mucinous tumours may show little uptake ( Fig. 22.24 ).

CT estimates the T stage, with accuracy for identification of disease extension beyond the bowel wall of around 86%. Colonic segments with a mesentery are fully enveloped by visceral peritoneum and hence tumours here are more likely to achieve T4a status ( Fig. 22.25 ). In the retroperitoneal colon, there may be direct retroperitoneal invasion, which may compromise the retroperitoneal surgical margin—a risk factor for local recurrence. Generally, CT is better suited to broad categorisation of tumours into good-prognosis and poor-prognosis groups. Such classification predicts outcome almost as well as histopathological analysis. This allows the targeting of poor-prognosis patients for neoadjuvant chemotherapy, as noted above, a topic under evaluation in clinical trials. Adverse prognostic features are unequivocal T3 tumours, T4 tumours and tumours with extramural venous invasion (EMVI). T3 disease should be suspected when there is extension of a discrete mass through the muscle coat into pericolic fat. EMVI is recognised by nodular or undulating expansion of colic veins.

CT performs poorly for nodal staging, as size thresholds are neither sensitive nor specific; over 50% of nodes shown to be involved at pathology measure less than 1 cm. Abnormal clustering of normal-sized nodes has been used as an alternative criterion but does not substantially improve performance. Ascites and omental or peritoneal nodularity indicate disseminated intraperitoneal disease.

Randomised trial data show that, in symptomatic adults, barium enema misses twice as many cancers as CTC and should be abandoned.

Rectal Cancer

Treatment of rectal cancer involves en bloc resection of the tumour, rectum and mesorectum (total mesorectal excision, TME) to minimise the risk of local recurrence. The dissection plane extends along the mesorectal fascia, constituting the circumferential resection margin (CRM). MRI is the imaging investigation of choice for local staging of rectal cancer. In particular, high resolution (1 mm ³ voxel size), T ₂ weighted (T ₂ W) fast spin-echo (SE) sequences perpendicular to the lesion provide information on local stage and relationship to the mesorectal fascia. Pre-operative MRI stratifies tumours into three main groups: (1) those highly likely to have an involved CRM after surgery; (2) intermediate-risk tumours that do not threaten the CRM; and (3) low-risk tumours. The first group usually require pre-operative downstaging (e.g. with chemoradiotherapy), to allow an attempt at curative surgery by TME; in the second group, operative therapy may be supplemented by neo-adjuvant therapy to reduce the risk of local recurrence; and the third group can undergo primary surgery with good local control rates without the toxicity of radiotherapy. CRM involvement is likely if MRI shows a tumour within 1 mm of the mesorectal fascia ( Fig. 22.26 ). Low rectal tumours that extend into the intersphincteric plane ( Fig. 22.27 ) are also high risk and may require neoadjuvant treatment and/or an extra-levator surgical approach to ensure clear resection margins.

Some early tumours (T ₁ and T ₂ ) can be treated by local resection alone. Transanal endoscopic microsurgery (TEM/TEMS) removes the cancer without the need for a full TME. Endorectal ultrasound (ERUS) is not able to resolve the mesorectal fascia, but has better differentiation of the wall layers than MRI and hence is utilised in some centres as an adjunct to establish local resectability of early-stage disease.

Nodal staging is important but challenging on MRI; mesorectal nodes will remain in situ with transanal microsurgical approaches and nodes outside the CRM will not be resected routinely at TME. Morphological features are more useful than size criteria. Irregular outline and heterogeneous internal signal are suspicious ( Fig. 22.28 ). Furthermore, diffusion-weighted imaging (DWI) shows some promise for nodal staging. Even more encouraging is the performance of contrast-enhanced MRI using blood-pool agents, although both these and iron oxide MRI contrast agents have limited availability, reducing clinical uptake.

EMVI is the direct invasion of a blood vessel by tumour. It can be detected on MRI as abnormal signal intensity expansion of perirectal vessels ( Fig. 22.29 ) and can be seen in conjunction with discontinuous vascular tumour deposits ( Fig. 22.30 ). EMVI is an important prognostic marker associated with haematogenous spread and hence its radiological identification and quantification may influence treatment by, for example, early addition of systemic chemotherapy.

Tumour response evaluation ( Table 22.4 ) after chemoradiotherapy is important; a proportion of patients have a complete response with no viable tumour at pathological examination after resection ( Fig. 22.31 ). MRI, in combination with clinical and endoscopic assessment, may be able to identify the patients who could forgo surgery and enter a ‘watchful waiting’ surveillance regime, particularly when DWI sequences are used ( Fig. 22.32 ). This remains under investigation, and is not yet standard practice, but has a role in selected cases with particularly high operative risk.

TABLE 22.4

Tumour Response Assessment on Magnetic Resonance Imaging

MRI Tumour Regression Grade (MR-TRG)	Response	Imaging Features
TRG 1	Complete radiological response	No evidence of tumour signal or dense fibrosis only
TRG 2	Good response	Dense fibrotic scar with minimal evidence of residual viable tumour
TRG 3	Moderate response	Mixed fibrosis and/or mucinous degradation with intermediate signal representing residual tumour but fibrosis predominates
TRG 4	Slight response	Partial response. Fibrosis or mucinous degradation is present but residual tumour predominates
TRG 5	No response	No evidence of response to therapy—tumour has same appearance as baseline

MRI , Magnetic resonance imaging.

MRI is also useful for imaging recurrent cancer, although FDG-PET/CT is probably superior especially as it can simultaneously exclude distant disease that may be inconspicuous on conventional imaging. The key roles for MR are to determine suitability for resection and to guide the operative strategy (e.g. hemiclearance vs. total pelvic exenteration) and the need for associated vascular reconstruction or sacral resection.

Anal Cancer

This rare tumour is increasing in prevalence and is probably related to an increase in human papilloma virus (HPV), which is a leading aetiological factor. Squamous histology is typical. Tumours with an epicentre caudal to a point 2 cm above the dentate line should be staged as anal tumours. MRI is the test of choice for local staging, although endoanal ultrasound (US) imaging may have a role. The T stage ( Table 22.5 ) depends primarily on size and invasion of local organs (see Fig. 22.33 ). Nodal drainage is usually to the inguinal region for tumours below the dentate line and the perirectal and internal iliac nodes for those above it. Metastases are best shown with FDG-PET/CT, although conventional CT is an alternative. MRI after standard chemoradiotherapy treatment can detect residual or recurrent tumour and facilitate selection for salvage surgery (abdominoperineal excision) ( Fig. 22.34 ).

TABLE 22.5

Anal Cancer Staging (TNM 8th Edition)

UICC/TNM	Tumour Extent	–5 year survival Squamous cancers	–5 year survival Non-squamous cancers
Stage I	T1—Tumour not more that 2 cm in greatest dimension N0—No regional lymph node metastasis M0—No distant metastasis	77	71
Stage IIa	T2—Tumour more than 2 cm but not more than 5 cm in greatest dimension N0—No regional lymph node metastasis M0—No distant metastasis	67	59
Stage IIb	T3—Tumour more than 5 cm in greatest dimension N0—No regional lymph node metastasis M0—No distant metastasis
Stage IIIa	T1 or T2 N1-Metastasis in regional lymph node(s) N1a Metastases in inguinal, mesorectal, and/or internal iliac nodes N1b Metastases in external iliac nodes N1c Metastases in external iliac and in inguinal, mesorectal and/or internal iliac nodes M0—No distant metastases	51	35
Stage IIIb	T4—Tumour of any size invades adjacent organ(s) (e.g. vagina, urethra, bladder); direct invasion of the rectal wall, perirectal skin, subcutaneous tissues or the sphincter muscles is not classified as T4 N0—No regional lymph node metastasis M0—No distant metastasis
Stage IIIc	T3 N1-Metastasis in regional lymph node(s) M0—No distant metastasis
Stage IV	Any T Any N M1	15	7

TNM, Tumour node metastasis; UICC, Union for International Cancer Control.

Appendix Tumours

Around 1% of appendix specimens contain neoplasia, usually carcinoid or adenocarcinoma. Carcinoids are often small, incidentally detected, and adequately treated by the appendicectomy that led to the diagnosis. Conversely, mucinous adenocarcinomas may be problematic. Residual mucin can lead to pseudomyxoma peritonei or retroperitonei ( Fig. 22.35 ).

Although considerably less common than appendicitis, mucocoele of the appendix ( Fig. 22.36 ) is associated with increased risk of malignancy with histological evidence of cystadeno-carcinoma in approximately one-third of cases. Furthermore, appendix mucocoele is associated with synchronous colonic tumours and hence colonoscopy is recommended.

Summary Box: CT and MRI for Diagnosis and Staging of Large Bowel Malignancy

• Sensitivity of computed tomography (CT) colonography for detecting colorectal cancer and large polyps approaches that of colonoscopy when performed using optimal technique and interpreted by appropriately trained practitioners.

• CT remains the mainstay for staging distant disease from large bowel primary tumours, although positron emission tomography combined with either CT or magnetic resonance imaging (MRI) has increasing utility in identifying occult metastases in high-risk patients.

• MRI is the technique of choice for local staging of anal and rectal tumours evaluating stage, adverse prognostic features and circumferential resection margin status with high accuracy, providing an invaluable tool for guiding surgery or neoadjuvant therapy.

• MRI can also demonstrate tumour regression following neoadjuvant therapy helping to inform clinical decision making non-operative approaches to rectal malignancy.

• Nodal staging for colonic malignancy remains challenging for all imaging techniques; morphological characteristics should be considered rather than size criteria alone to assess likelihood of involvement.