The accuracy and reproducibility of preoperative MRI in identifying prognostic risk factors such as depth of invasion and safety of the surgical circumferential resection margin are accepted [1, 2]. MRI-based staging is increasingly recommended [3] because patients at highest risk can undergo preoperative adjuvant treatment, which has a better outcome and less toxicity than does postoperative treatment [4].

Histologic extramural vascular invasion has long been recognized as an independent predictor of local and distant recurrence and poorer overall survival [5–7]. It is defined as the presence of malignant cells within endothelial cell–lined blood vessels beyond the muscularis propria [8] and is reported to occur in as many as 52% of cases of colorectal cancer [5, 8–10]. It is possible to detect extramural vascular invasion with MRI [11]. It has been found that the severity of MRI-detected extramural vascular invasion correlates with relapse-free survival [12] and may be important when preoperative treatment strategies are being considered for patients with this particularly poor prognostic feature. To our knowledge, the radiologic characteristics of extramural vascular invasion have not been adequately described in the literature. This article illustrates the imaging criteria associated with extramural vascular invasion by showing the correlation between findings on MR images and in whole-mount pathologic sections. The grades of severity important in the preoperative assessment of extramural vascular invasion are depicted.

High-spatial-resolution MRI facilitates visualization of vascular structures in situ within the mesorectum over a series of adjacent images. Veins may be recognizable on T2-weighted images as serpiginous or tortuous linear structures [11]. Larger vessels may appear black owing to signal void, and smaller vessels may be recognized because of tortuosity and branching (Fig. 1). Very small unnamed vessels may radiate outward from the edge of the muscularis propria into the perirectal fat (Fig. 2). Larger named vessels, such as the superior rectal vein and middle rectal vein, may be visualized in a consistent anatomic position, and a contralateral paired vessel may be present, helping with identification (Figs. 3 and 4). It is not always possible, however, to determine with absolute certainty whether a structure is vascular, and this lack of clarity is an important limitation in the radiologic assessment of extramural vascular invasion.

Depth of tumor invasion can indicate the potential for extramural vascular invasion. By definition, histologically defined extramural vascular invasion must be associated with tumors that are at least category T3. Therefore, a small tumor clearly limited to within the muscularis propria (category T2 or T1) has no potential for invading extramural vessels. Assessment of MR images for features suggestive of extramural vascular invasion must include the following four components: pattern of tumor margin, location of tumor relative to major vessels, caliber of vessel, and vessel border.

Radiologically, the tumor margin may appear nodular or smooth. Tumor invasion into the small noncharacterizable veins that radiate outward from the bowel wall gives rise to a nodular border (Fig. 5A, 5B). This finding can be differentiated from desmoplasia, which appears as fine stranding of low signal intensity (Fig. 6A, 6B, 6C).

Whenever tumor is seen to lie close to a vessel, the radiologist should consider the possibility of extramural vascular invasion (Fig. 7A, 7B). Histologically, vascular invasion occurs when the tumor directly penetrates a vessel wall before extending along the lumen. Therefore, the presence of tumor signal intensity within a vascular structure is highly suggestive of extramural vascular invasion.

As tumor invades along the lumen, the vessel expands. Tumor signal intensity is intermediate (gray), and therefore any expansion of a low-signal-intensity vessel by tumor invasion usually is identifiable. The caliber or size of any vessels seen on MR images can be described as normal, slightly expanded (Fig. 8), or grossly expanded (Fig. 9). Eventually, the tumor may expand through and beyond the vessel wall, disrupting the border, which can be described as either smooth (normal) or irregular or nodular (Fig. 9).

The degree of extramural vascular invasion visible at MRI ranges from very minimal to very extensive. According to the four criteria (tumor margin, tumor location relative to vessels, vessel size, and vessel border), a 5-point grading system for the MRI-based preoperative assessment of extramural vascular invasion has been proposed [12]. This MRI extramural vascular invasion score is logical. The lowest score, 0, correlates with the absence of any feature suggestive of extramural vascular invasion. The maximum score, 4, is given when the most overt features (grossly expanded and irregular vessel border) are seen. It is helpful for radiologists to be able to score MRI extramural vascular invasion because it has been shown that higher scores are associated with poor survival. Lower scores are not associated with histologic extramural vascular invasion or with adverse outcome [12]. Stratification of patients into prognostic groups according to MRI extramural vascular invasion score is clinically relevant both for preoperative treatment of patients at high risk and for postoperative follow-up.

Because of limitations of resolution, microscopic examples of extramural vascular invasion are not detectable with MRI (Fig. 7A, 7B). Patients with large-vessel invasion have the worst outcome [8]. Therefore, the clinical significance of extramural vascular invasion not detectable on MR images may be minimal [12]. In contrast, extensive vascular invasion can destroy the vessel wall, leaving little evidence of normal venous cellular architecture, and such cases may be underreported by pathologists. The main advantage of using contiguous 3-mm slices for MRI is the ability to ascertain that tumor actually lies within a vessel, either because the vessel is seen in the typical anatomic position of one of the main vascular structures, such as the middle rectal vein (Fig. 10A, 10B, 10C, 10D), or because a normal, unexpanded vein is seen extending beyond the area of tumor signal intensity in the same or an adjacent image slice (Fig. 11A, 11B, 11C, 11D).

Extramural vascular invasion is an important prognostic factor in colorectal cancer. Recognition of invasion is important in the preoperative staging of rectal tumors. The criteria used for MRI extramural vascular invasion scoring are based on pathologic and anatomic considerations, and the scoring system is straightforward and logical. Prognostic stratification based on MRI extramural vascular invasion score has been found to correlate with clinical outcome. In addition, MRI can be seen to contribute additional staging information, depicting gross vascular invasion that is not necessarily recognized histologically.

We thank M. Elmahallawy for preparing the Van Gieson stain and Barbara Bannerman for assistance in the preparation of the manuscript. We thank the radiologists, pathologists, surgeons, and nurse specialists who were members of the MERCURY Study Group and were responsible for the conception and design of the study and collection of data in the original MERCURY study. The following hospitals took part in MERCURY: Pelican Cancer Foundation, The Ark, North Hampshire Hospital; Royal Marsden Hospital Colorectal Network, Royal Marsden Hospital; Epsom and St. Helier NHS Trust; Mayday University Hospital; North Hampshire Hospital; Leeds Hospitals Teaching Hospitals, Leeds General Infirmary; St. James's University Hospital; Norwegian Radium Hospital; Frimley Park Hospital; Ashford and St. Peter's NHS Trust; Krankenhaus im Friedrichshain; Llandough Hospital; Karolinska University Institute.

Presented as a poster at the 2007 annual meeting of the Association of Coloproctology of Great Britain and Ireland, Glasgow, Scotland.

N. Smith's research post was funded by the Croydon Colorectal Cancer Charity, which had no involvement in the study design; in the collection, analysis, or interpretation of the data; or in the writing of the report.

Address correspondence to G. Brown ([email protected]).